Regulation of Artificial Intelligence Bill 2026: Second Stage

Minister of State at the Department of Enterprise, Tourism and Employment (Deputy Niamh Smyth)

I move: "That the Bill be now read a Second Time."

I am very pleased to bring this Bill before the House today. The purpose of the Bill is to give full effect in Ireland to EU Regulation 2024/1689 of the European Parliament and Council, and of the EU's Artificial Intelligence Act, which I will refer to throughout as the AI Act.

The Al Act is designed to provide a high level of protection to people's health, safety and fundamental rights and to promote the adoption of human-centric, trustworthy Al.

As with all EU regulations, the Al Act has direct legal effect across all member states. It aims to harmonise rules for Al systems across the EU, creating a single market and ensuring a consistent approach to risk, safety and fundamental rights. With those aims in mind, this Bill is a technical implementing measure. It neither adds to nor alters the obligations placed on regulated entities by the Al Act. The Bill establishes essential domestic regulatory architecture for the supervision and enforcement of the Al Act in the State.

A central feature of the Bill is the establishment of oifig intleacht shaorga na hÉireann, the Al office of Ireland, as an independent statutory body. Our new Al office will be designated as the single point of contact for our citizens at home, the European Commission and other member states. In this role it will support consistent enforcement of the Act and facilitate co-operation between competent authorities and fundamental rights bodies. The office will promote Al innovation and literacy and oversee the Al regulatory sandbox, which will foster innovation and support startups and SMEs in compliance with the Al Act.

The Bill also empowers our competent authorities and it creates a robust administrative sanctions framework. The Bill establishes a regulatory foundation on which to build trust in society for an evolution of Al that serves as a tool for people, with the ultimate aim of increasing human well-being.

Artificial intelligence is one of the defining technologies of our era. Its applications now reach into virtually every corner of the economy and society, from healthcare and financial services to the delivery of public services and much more. Al has brought remarkable benefits. It is accelerating scientific discovery, improving outcomes in medicine, expanding access to services and generating significant economic opportunity. Our national digital and Al strategy, Digital Ireland: Connecting our People, Securing our Future, sets out Ireland's ambition to be a global hub for applied Al innovation.

Al also carries real risk, however. Without appropriate oversight, it can embed and amplify discrimination, manipulate behaviour or exploit the vulnerabilities of people, posing threats to the health, safety and fundamental rights of our citizens. The Al Act, which entered into force in August 2024, is the first ever legal framework designed to address these concerns while simultaneously fostering innovation of responsible and ethical Al. The Al Act is a risk-based, horizontal regulation. It applies across all sectors of the economy, both private and public, with targeted exemptions for national defence, national security and purely personal use. Crucially, its measures are proportionate: most Al systems will give rise only to minimal risk and can be marketed and used subject to existing legislation, without additional obligations under the Al Act. This ensures that the EU and Ireland remain competitive for Al investment and innovation. The Al Act also takes into account the specific interests and needs of SMEs in terms of simplified obligations and tailored supports.

There are five main pillars to the Al Act, which set out obligations across a number of categories and uses of Al systems and models. Under prohibited practices, eight Al practices are prohibited due to the unacceptable risk they pose to European values and fundamental rights, including practices such as harmful manipulation, social scoring and real-time remote biometric identification. Given the rapid development of technology, the Al Act also provides for the list of prohibited practices to be kept under regular review. Stringent requirements apply to high-risk Al systems used in eight areas set out in the Al Act, such as biometrics and critical infrastructure. Under transparency obligations, specific requirements apply to Al systems that interact with users, generate synthetic content with risks of impersonation or enable deepfakes. Under the general purpose Al pillar, providers of general-purpose Al models face obligations to mitigate systemic risks. Enforcement will be led by the European Commission with member state co-operation. Under the enforcement pillar, member states are required to designate competent authorities to supervise and enforce the regulation of Al and to provide for significant penalties, with fines of up to €35 million or 7% of global annual turnover for the most serious infringements.

I will now speak to the importance and urgency of the legislation. Ireland has a key position in the European regulatory landscape. Eight of the leading providers of foundational Al models are established here. These are the companies at the frontier of Al development and deployment globally. Maintaining the competitiveness and viability of Ireland's digital economy is an overarching national imperative, as outlined in Programme for Government: Securing Ireland's Future, and is the driver behind a number of key commitments in the Government's recently updated national Al and digital strategy.

As Ireland assumes the Presidency of the Council of the EU, meeting our obligations under the Al Act will further strengthen our position as an EU centre of excellence and digital regulatory hub. It will also demonstrate our commitment to fostering responsible, human-centric Al innovation, both domestically and across Europe.

There is also a firm legal deadline. Enforcement of prohibited Al practices posing unacceptable risks, as set out in Article 5 of the Al Act, takes effect as of 2 August 2026. Ireland must have the necessary national supervisory and enforcement infrastructure in place by that date. I acknowledge the considerable complexity and pace of work required to reach that point. The EU allowed a comparatively short window between the Al Act entering into force and its application. The complexity of the Bill, combined with the parallel development of the EU's digital omnibus proposal, has required extraordinary efforts from my officials, the Office of the Parliamentary Counsel and the Office of the Attorney General. To them I am most grateful.

As I have noted, and as Deputies will no doubt be aware, the EU's digital omnibus on Al, a set of targeted amendments to the Al Act, was provisionally agreed in May 2026 and approved by the European Parliament in June. Ireland supports the agreed omnibus, which helps increase legal certainty and reduce administrative burden. Importantly, the omnibus extends certain implementation timelines, including for transparency obligations and high-risk Al system requirements. However, the omnibus does not extend the timeline for enforcement of prohibited Al practices under Article 5. The 2 August 2026 deadline remains firm, and this Bill addresses it directly. Any amendments to domestic legislation that may be required following formal adoption of the digital omnibus on Al will be addressed as part of phase 2 of this work in the autumn. The second phase will also provide an opportunity to further enhance Al regulation in accordance with the Al Act if required.

Having set out the context and the urgency of the Bill, I will now outline its provisions. The Bill is structured in ten Parts, with 139 sections and 4 Schedules. Its overarching policy objective is to ensure that the Al Act can be comprehensively and effectively implemented here in Ireland.

Part 1 of the Bill, comprising sections 1 to 5, inclusive, provides for the Title and commencement, key definitions, regulation-making powers, service of documents, and expenses, to provide that the costs of administering the Act are funded with the appropriate ministerial approval.

Part 2 is set out over sections 6 to 40, inclusive, and provides for the establishment of oifig IS na hÉireann and its governance structure, including appointment of its board, CEO and staff and its accountability, funding and functions. Sections 12 to 23, inclusive, are concerned specifically with matters relating to board governance, with provisions on board establishment and membership, appointment process, terms of appointment, membership of either House of the Oireachtas, remuneration, conflict of interest and others. Sections 24 to 33, inclusive, relate to the chief executive officer role and provide for the matters relating to the CEO's appointment, resignation, accountability and functions. Sections 34 to 36, inclusive, relate to the staff of oifig IS na hÉireann and confidentiality.

The appointment of staff is subject to the Public Services Management (Recruitment and Appointments) Act 2004 and the Civil Service Regulation Acts 1956 to 2005. Sections 37 to 40, inclusive, provide for funding and accountability, including the strategy statement, accounts and annual report.

Part 3, comprising sections 41 to 52, inclusive, provides for the practical operational infrastructure underpinning enforcement of the Al Act in Ireland. Section 42 designates the Al office as the national single point of contact in relation to the public and other counterparts at member state and Union levels, and to help increase organisational efficiency as required by the Al Act. Section 43 provides that the Al office will establish and maintain a national Al register, recording all instances of prohibited Al practices, serious incidents involving high-risk Al systems, and other required notifications. Transparency and public accountability are embedded from the outset. Section 46 provides for the establishment of Al regulatory sandboxes under the supervision of the Al office. SMEs and start-ups are to receive priority access. Sections 48 to 50, inclusive, make provision for real-world testing of high-risk Al systems in controlled conditions, subject to appropriate safeguards, oversight, and rights of review.

Market surveillance authorities are a key cohort of the national competent authorities required for the implementation of the Al Act and Part 4 sets out, at sections 54 to 62, inclusive, the general obligations and powers of market surveillance authorities, including the power to receive complaints from the public. Section 59 enables fundamental rights bodies to request the testing of high-risk Al systems where infringements of fundamental rights are suspected. Sections 63 and 64 are concerned with co-operation agreements of relevant market surveillance authorities. Sections 65 and 66 set out data protection matters for relevant competent authorities.

Part 5 provides market surveillance authorities with a comprehensive, structured, and proportionate enforcement toolkit, progressing from co-operative compliance measures through to coercive intervention. Authorised officers, appointed by each market surveillance authority under sections 68 to 70, inclusive, will have powers to enter premises, inspect relevant Al products and records, require the production of information, and take samples. Section 71 provides that contravention notices may be issued requiring an operator to remedy a breach. Section 72 provides that prohibition notices may be issued to stop an Al system from being placed on the market or put into service where it poses a risk or breaches the Al Act. Sections 73 and 74 provide powers for seizure, disposal, and forfeiture of unsafe Al products and, under section 75, in the most serious cases, for the removal of online content to eliminate serious risk. Section 77 provides rights of appeal to the District Court throughout.

Part 6 establishes, at sections 78 to 120, inclusive, the independent adjudication system for the imposition of administrative fines for breaches of the Al Act. Sections 79 to 86, inclusive, provide that where a market surveillance authority believes an infringement has occurred and an authorised officer has conducted an investigation, the matter may, with the authority's consent, be referred to an independent adjudicator. The administrative fines, set out in section 105, are substantial and fully aligned with fines set out under the Al Act. For public bodies, fines are capped at €1 million, in line with the data protection legislation. Section 114 provides that no administrative fine takes effect without confirmation by the High Court, which is a key procedural safeguard. Sections 107 to 109, inclusive, provide for the rights of appeal. Sections 116 to 117 provide for the nomination of adjudicators by the Al office and appointment by the Minister for Enterprise, Tourism and Employment. Section 118 expressly provides for the adjudicators’ independence in the performance of their functions.

Part 7 of the Bill concerns penalties and miscellaneous provisions and, at sections 121 to 127, inclusive, sets out criminal penalties for obstruction of authorised officers, failure to comply with enforcement notices, and unauthorised disclosure of confidential information. Part 8, at sections 128 to 133, inclusive, amends the Central Bank Act 1942 to equip the Central Bank with the investigatory and sanctioning tools necessary for its role as a market surveillance authority under the Al Act. Part 9, at sections 134 to 137, inclusive, amends the Competition and Consumer Protection Act 2014 to equip the Competition and Consumer Protection Commission with the investigatory and sanctioning tools necessary for its role as a market surveillance authority under the Al Act.

Part 10, at sections 138 and 139, contains miscellaneous amendments, including consequential amendments to the Communications Regulation Act 2002, and the Freedom of Information Act 2014. Additionally, there are four Schedules governing oral hearing procedures, the assignment of adjudicators and the register of adjudicators.

In conclusion, establishing a credible and coherent Al enforcement regime in Ireland underscores our commitment to responsible Al development and is a core component of our international competitiveness offering. In this respect, the Regulation of Artificial Intelligence Bill 2026 matters to how Al will impact our society, our economy, our international standing and, of course, our future. It puts in place the enforcement structures that will protect people in Ireland from Al systems that pose unacceptable risks to their health, safety, and their fundamental rights. It provides the regulatory clarity and helps build the public trust that businesses, both indigenous and multinational, need to invest, operate, and grow with confidence.

As we assume the Presidency of the Council of EU, meeting our obligations under the Al Act demonstrates our commitment to responsible digital governance and to being a trusted partner in the EU. The implementing measures enabled by this legislation, including the establishment of the new Al office of Ireland, will provide an enduring foundation, strengthening our position as an EU centre of excellence and digital regulatory hub, and demonstrate our commitment to fostering and promoting responsible Al innovation and adoption both in Ireland and in Europe. It is imperative that the State has a comprehensive and robust legal basis for the full and effective implementation of the Al Act. I am confident this Bill achieves that objective in a balanced, proportionate and technically sound manner. I thank colleagues for their attention and commend the Bill to the House.

Deputy Rose Conway-Walsh

Artificial intelligence is transforming our economy, workplaces and daily lives at an extraordinary pace. It offers enormous opportunities to improve public services, drive innovation and enhance productivity but it also presents serious risks that cannot be ignored. We all can recognise the benefits AI can bring to individuals, business and the public sector, be it through task automation, improved data analysis or increased efficiency and competitiveness across many sectors. However, AI is not an inherently positive force. It is a tool that can be used for good or bad, depending on the intentions of those who develop and deploy it. This is precisely why it must be properly and robustly regulated.

I welcome the Regulation of Artificial Intelligence Bill 2026, which will establish the national structures required to implement and enforce the EU AI Act in Ireland. It will provide the governance, supervision and enforcement framework necessary to ensure Ireland meets its obligation to have a fully operational AI regulatory regime in place by 2 August 2026. This important deadline falls during Ireland's Presidency of the European Union. As a country that has positioned itself as a major digital hub within Europe, it is important that Ireland demonstrates leadership and a clear commitment to online safety, accountability and the responsible use of emerging technologies.

The establishment of a dedicated national AI office is indeed welcome. However, its effectiveness will depend on whether it has the necessary independence, staffing and financial resources to carry out its functions properly.

It will be important that the board members of the new office collectively possess expertise in AI and computer science, human rights and equality law, trade union and workers' rights matters, consumer protection and civil society engagement. There must be no conflict between the State's deployment of AI and its role in promoting innovation and its responsibility to regulate and enforce compliance. The office must have sufficient authority and capacity to support other regulators and fundamental rights bodies. We in Sinn Féin support empowering enforcement bodies to tackle the misuse of artificial intelligence and protect citizens from harmful or unlawful practices. However, enforcement mechanisms must be matched by adequate resources, transparency and oversight.

While much of the debate around artificial intelligence focuses on innovation and competitiveness, we must not lose sight of its impact on workers. Technological advancement should improve people's lives, not undermine their livelihoods. AI should enhance human labour, not replace it. We are already witnessing the consequences of AI-driven restructuring. Recently, 720 workers in Covalen learned their jobs are to be made redundant as advances in artificial intelligence increasingly replace functions previously carried out by people. The Government cannot simply sit back and hope for the best as artificial intelligence reshapes our labour market.

A recent report by the ERSI and Department of Finance estimated that around 7% of jobs in Ireland could be displaced by AI in the short- to medium-term, the equivalent of almost 200,000 jobs. This should serve as a wake-up call that future displaced workers need meaningful protections and supports to be put in place. The new AI office must promote the development and deployment of AI systems that advance high-quality employment, regional development, social inclusion and the provision of public services. Workers must be supported to adapt to changing technologies, gain new qualifications and transition into emerging sectors. More broadly, I want the Government to develop a comprehensive national strategy to manage the impact of AI on employment and ensure workers are partners in shaping technological change, rather than casualties of it. The new AI office could contribute significantly to this work.

The Government has already stated that the Bill does not create substantive administrative obligations beyond those already contained in the EU AI Act. With this Bill we need to be careful not to over-regulate AI and add to the already large administrative burden of SMEs. It is a difference between the administrative burden and regulation that we need. We must also ensure Irish companies can access AI technology to the same extent as international competitors, otherwise, Ireland's competitiveness will be negatively impacted. Regulation must protect citizens while also enabling responsible technological development and economic growth. For Sinn Féin, regulation of AI must be guided by clear principles. Transparency is essential. AI-generated content and tools should be clearly identified where there is potential for public consumption. Copyright protections must be upheld to protect artists, designers, writers and those working in creative industries from having their original works replicated and exploited without consent.

Privacy rights cannot be compromised. AI systems must comply fully with the GDPR and all relevant data protection legislation. Personal information, including sensitive and biometric data, must be processed lawfully, fairly and transparently. Fairness is another key consideration. AI must never reinforce discrimination, inequality or social exclusion. I say that in particular in regard to access to vital public services by elderly people, people with disabilities and vulnerable people in our communities. Regular monitoring of systems is required to identify and eliminate biases, in particular where AI is used in employment decisions, recruitment processes or the delivery of public services.

We must also recognise the growing cybersecurity threats associated with artificial intelligence. AI has already been used by malicious actors to create increasingly sophisticated phishing exercises, deepfake scams and fraudulent communications. It can automate cyberattacks, identify vulnerabilities at scale and facilitate the spread of misinformation. As AI capabilities become more advanced, the potential for criminal exploitation also increases. The State must ensure that cybersecurity considerations are embedded in AI regulation and governance frameworks. Businesses, public bodies and critical infrastructure providers must be equipped to deal with these evolving threats. Therefore, public engagement and education are vitally important. The Government must be proactive in informing citizens about what artificial intelligence is, how it is being used and what risks and opportunities it presents. Public awareness campaigns, online tutorials and our education system all have a role to play in ensuring people can use these technologies safely and competently.

Particular attention must be paid to vulnerable groups in society, as I said. The rights of children and the elderly must come first and we must have meaningful engagement with people with disabilities to better understand their experiences with AI technologies. Artificial intelligence is evolving much faster than legislation. The Government must remain agile and capable of responding quickly to emerging risks and developments. Sinn Féin supports the principle of effective AI regulation. We welcome measures that protect citizens, uphold fundamental rights and ensure accountability, but regulation alone is not enough. The Government must ensure that workers are protected, citizens are informed, enforcement bodies are adequately resourced and innovation proceeds in a way that benefits society as a whole. Artificial intelligence should serve people, strengthen our communities and support our economy, while always remaining subject to democratic oversight, human judgment and public interest.

On SMEs, in particular small businesses and the use of AI, we see the use of AI in big tech. We need to be conscious of the supports that are necessary for small and medium businesses, including small retailers, in order to be able to avail of the opportunities of AI while also protecting their businesses. I want a front-loading of the supports and services that are available to SMEs.

Deputy Darren O'Rourke

I welcome the opportunity to speak on the Regulation of Artificial Intelligence Bill 2026. It represents a significant step in establishing the institutional and enforcement framework for the EU AI Act in Ireland. It creates a new AI regulator as an independent statutory body and establishes the supervisory powers and penalty regime we need to meet our obligation to have an operational regulatory framework in place by 2 August. That deadline falls during our EU Presidency. As a digital regulatory hub, it is essential that we demonstrate our commitment to online safety.

Let me be clear from the outset. The Bill is the absolute minimum required. Meeting the deadline is not the same as meeting the moment. I ask that the Minister of State engage with the Opposition on the Bill as it goes through the Houses because there are opportunities for improvement to every aspect of it, to put it mildly. Fundamentally, I would argue that it prioritises economic competitiveness over the protection of fundamental rights. It is important that we get the Bill right.

There are systemic gaps in the regulatory architecture. The enforcement regime is not what it needs to be. The AI office needs to be more than a single point of contact. I know from different iterations of the Bill and pre-legislative scrutiny that some surveillance powers or duties were not included.

Regarding the enforcement regime and its funding and resourcing, there is a deep concern that we will end up with a regime that is incredibly fragmented and impossible to navigate. There are also very real concerns in relation to child protection. Children are a vulnerable group and there is concern to ensure their fundamental rights are protected.

It was frustrating for the members of the AI committee that we were not more central to the design of this legislation. However, we have an opportunity now to discuss it. Concerns were raised about the independence of the AI office and how it will be constituted. I am open to discussion on this but I still question the influence of the Minister, the role of the Department and the potential for executive overreach. Along with my colleagues in Sinn Féin, we will chart our own way forward on this legislation but I am flagging these issues at this Stage because they are concerns that immediately are apparent to me and I know others have raised, and will raise, similar concerns.

We in Sinn Féin recognise the immense potential of artificial intelligence. The benefits are there to see, including automation of repetitive tasks, improved accuracy in data analytics, enhanced healthcare outcomes and increased productivity. The potential at least is there for those outcomes to be achieved. However, we are equally, if not more, aware of AI's profound capacity for harm. Those aspects are very apparent and have already been seen. This is not a choice between embracing the future and protecting our citizens. It is our duty to do both and to do so with foresight and determination.

The Grok scandal some months ago, where generative AI was used to create non-consensual sexualised deepfake images of thousands of Irish women and children, was a watershed moment. It was not an event that could not be predicated. Rather, it was a direct consequence of a regulatory vacuum and the failure of successive Governments to act with the vigilance and urgency the situation demanded. The scandal exposed a profound failure to protect citizens from the predatory usage of AI. What happened was not a glitch or accident and it certainly was not acceptable.

Sinn Féin brought forward legislation to close the loophole in Coco's Law by explicitly criminalising the generation of non-consensual intimate deepfake imagery. While the Government did not oppose our Bill, it remains to be seen whether it will match our and others' commitment to this measure with meaningful action. The message must be made clear that creating such material is illegal. A zero-tolerance approach to violence against women and children means taking action when new threats emerge, not waiting for harm to be done before we intervene. The victims of such attacks deserve more than sympathy. They deserve justice. The State's approach to AI regulation cannot simply be reactive. The Government cannot afford to wait for a scandal to force its hand.

For Sinn Féin, the focus is clear. Safety, ethics and fundamental rights must be central to the design of our regulatory framework. We must ensure that technological advancement does not come at the expense of the rights and dignity of our citizens, particularly our most vulnerable and especially children. When we talk about vulnerable groups, we are talking about real people, including young girls whose images are stolen and manipulated, women whose career and reputation can be destroyed by a single malicious deepfake image and members of minority communities that are already disproportionately targeted by disinformation and algorithmic bias. These are not abstract risks. They are present dangers that require an immediate and robust response.

The Bill before us introduces enforcement powers. It will allow regulators to issue contravention notices requiring compliance, withdrawal, recall or destruction of AI systems. There is a need for improvement and enhancement of every aspect of the Bill. The Oireachtas AI committee has argued that the EU AI Act must be treated as the minimum legislative requirement, not the ceiling. As AI development moves much faster than legislation, the Government must be able to combat negative developments in a swift manner. We need regular monitoring of AI systems to ensure illegal activity, inequality, bias and discrimination can be detected and removed. That will protect citizens' rights where AI is used in decision-making, where deepfake images are circulated and when AI is used for the delivery of public services. It will ensure that racial, gender or socio-economic discrimination is prevented.

GDPR and copyright legislation must continue to apply to AI. New and stronger legislation may be necessary to ensure that AI systems process data in a lawful, fair and transparent manner. There must be proactive delivery of public information and education. We need to ensure AI is inclusive and accessible for everyone. Public campaigns, online tutorials and the education system must inform citizens about how to use AI in a safe and appropriate manner. The rights of children and young people need to come first. Operators must ensure their platforms account for child safety and protection. There must be engagement with people with disabilities and supports for older people navigating technological developments.

Adoption of AI comes with the danger of job losses, which we are already seeing in manufacturing, technology and finance. Sinn Féin advocates for strong protections to ensure technological advancements do not erode workers' rights, safety or employment security.

We must also look at the bigger picture. Sinn Féin has serious concerns about the direction of travel at EU level. The digital omnibus package significantly weakens broader AI regulations and I am proud to say our MEPs voted accordingly. We will not accept a future where safeguards are traded away in the name of simplification and where the primary focus is on facilitating innovation, profits or competition between tech bros to the neglect of fundamental rights. The drive for competitiveness cannot come at the cost of our citizens' safety. The Government must stand firm against any efforts to dilute protections. There are real questions in that regard. This Bill is an important step but it is really important to get it right. I appeal to the Minister of State to work with the Opposition to ensure that happens.

Deputy Sinéad Gibney

Cuirim fáilte roimh an mBille seo. Tá sé thar a bheith tábhachtach go mbeidh na dlíthe cuí againn chun dul i ngleic le AI agus go mbeimid réidh le déileáil leis an teicneolaíocht atá romhainn.

On behalf of the Social Democrats, I welcome the Bill and the discussion on Second Stage on the regulation of AI, as provided for in the Bill. Before speaking more specifically, including on some of the amendments I have already submitted, I will comment more broadly on AI and how it is impacting on our society. That is the basis on which we are trying to regulate its use. It is what is prompting governments and regulators around the world to deal with it.

My first point concerns the definition of AI. I am a member of the artificial intelligence committee, where we obviously discuss AI, and the communications committee, where part of our role involves the regulation of technology platforms. I am my party's spokesperson on enterprise, which is the area within which regulation of AI currently sits. In addition, data protection comes within the justice sphere. We are already in a very complex arrangement whereby a lot of different bodies and different perspectives are informing the discussion. I highlight this because in some people's mind, there is still a separation between technology in the form of online platforms and social media, for example, and AI. There is an idea that they are separate but that is no longer the case.

In fact, AI is the only show in town. It has long been the case that it is powering the social media platforms. The use of artificial intelligence is why algorithms are becoming more and more effective at making their products addictive. I worked in the technology sector between 2006 and 2014 and I remember someone telling me that some of the operations in the company I worked at went from a 5% level of automation to 95% during that period.

It is the case that AI is powering the technology all around us and it is not just the robots, the more exciting stuff that people sometimes associate with it. It is powering so much of the technology that reaches deep into our lives and it really has such a broad range. Some of it affects us directly, some of it is hidden, some of it is clearly technology and some of it we do not even associate with that. For me there are three different categories of concerns this raises.

The first one is the existential risk that we do not talk about it a lot in this Chamber. It has to be acknowledged that there are ways in which the technology is developing faster than not only regulators but even the engineers themselves can keep up with. We have seen headlines recently about this tipping point that we are currently getting to where AI agents can potentially start making decisions for themselves. They can start doing things which essentially contradict the wishes of the engineers who have developed that technology but are of a self-interest basis. That is the kind of dystopian world we see in movies and TV programmes and we do not really think it is ever going to happen. That is part of what is happening now and really why we are at such a critical moment right this minute of getting a handle on this and really getting control of it.

The second huge category that artificial intelligence raises for me as a concern is in labour displacement. Only this morning I was speaking to the Minister of State, Deputy Dillon, about the Covalen workers and the broader picture of AI labour displacement in Ireland. We are not dealing with this. In fact, it is for me the definition of "head in the sand" that when I raise this, the stock response from Government is that it knows AI is going to disrupt labour, but actually it has the potential to augment - I think that is the word the Minister of State, Deputy Dillon, used earlier - it as well, that it will improve jobs and we will have better-quality jobs and more critical thinking within our roles because all the automation and more mundane and tedious tasks will be done by technology. There is nothing in the path that has got us to this point so far that gives me confidence that this will be the case. It is automation which has transformed industries around us, like taxi drivers and delivery drivers more generally as well. None of it has been good for job security, job stability or for pay and conditions of those workers. What automation has done so far is to turn various parts of our employment sectors into a gig economy and it is not to the benefit of workers. It might be to the benefit of innovation, competitiveness and profit, but it is not to the benefit of workers. Unless we acknowledge that there is going to be major labour displacement as a result of artificial intelligence and unless we are honest with ourselves as policy makers and the Government is honest with itself, then this is going to be an issue that will absolutely dominate the next ten years of politics.

Within that for me one of the most important parts is that we do not have the structures for workers' rights in this country which allow us to deal with a crisis of this size. This is the thing. It should not be a crisis. It should be that if labour is being displaced that actually we have more time as a species for leisure to actually get to do things and we get to value, for example, professions that are traditionally undervalued, like teaching or caring, which are things that cannot be automated and are predominantly done by women. That is an opportunity, but unless we acknowledge it and are honest about it, none of these things are going to happen. As is the case with so much of AI there is opportunity, but unless we are honest about the threats, we will not be able to harness that opportunity and I have major concerns about that.

The last of the three categories that I would see as being evoked in this is the harms one. Grok was an example of that. I know the Minister of State's response was to be horrified, and rightly so, at what happened earlier this year when that feature became available. However, the reality is that this feature still exists. People can access it on their phones in this country using the right technology if they wish to do so. People can still nudify images. They can still promote child sexual abuse material and image-based sexual abuse using the technology that is made available to them by a private company. Nobody has yet been prosecuted, no sanctions have been levied and nothing has actually happened. I know there is a range of different responses in terms of inquiries, investigations, prosecutions and so on. I keep being told that there is this overburdensome regulatory framework that people are trying to deal with and yet we have a clear scandal and this overly burdensome regulatory framework cannot seem to react to it in any level of real time. Maybe it will get there eventually but the reality is that we can still access these features. That is just an example of the whack-a-mole model the Government will be dealing with. Grok is one feature on one application on one platform. It abounds. We do not know yet; just like we did not anticipate nudification being the crisis that it became, we do not know where the next one is coming from. Adding to the digital omnibus, which is essentially what happened, the very specifics around nudification, is not enough, because it is not treating the fundamental issue around the business model of these tech companies and the control of power they currently have.

I agree with other speakers who have mentioned that, unfortunately, the view of this Government, which is absolutely mirrored from Brussels, is a very industry-focued lens for the governance of AI. Everything is done through the prism of business and industry, competitiveness, enterprise and innovation. Unfortunately, that is not going to work. Establishing our AI office in the Department of enterprise is, to me, the first mistake we are making in that lens being applied to the development of this governance. It does not balance effectively the rights and protections that we need to bake in to the governance of this technology. Every time I hear a Government spokesperson, they speak about those things and say that the Government has to balance it with the protections, but nothing that is being done is convincing me that this is actually happening. I am sorry but it is just not coming through in the Government's actions.

I mentioned power and this is really important. It is becoming more and more accepted that the control and power of this technology is in the hands of too few people and specifically in the hands of five or six tech broligarchs in the States. This group of people is pushing Donald Trump to lobby Brussels to deregulate so that they can have more access to the market. We have heard them say this. They do not hide this. This control and power means that we will always face this imbalance. We will always have to deal with a heavy regulatory framework to try to wrestle back that control, but there are other ways we can do so. We need to open that conversation and be honest about it.

Another thing we need to think about is the just transition. We talk about a just transition in relation to the climate but we need to add to that a digital component. We are dealing with a transformation in society because of the man-made climate change and we are also dealing with a transformation in society because of the technology we are developing. Beyond the governance, we also need to talk about it at that very fundamental, societal level. Here is the thing; if we trust AI and we actually build governance that is effective and good, there is so much to harness. There is so much benefit that we can bring to society that we can all enjoy. I have said before that algorithms are not bad things. Algorithms are bad at the moment because they are designed to keep our eyes on the screen and to addict us to the products, but they are just a maths equation. If we instead were able to repurpose those algorithms to promote well-being, equality, employment, whatever it might be, whatever the social difficulty is that we are tackling at any given time within a jurisdiction, we can actually harness these tools for good social impact. We will never be able to do that as long as the control sits in the hands of so few people who are entirely motivated by profit. Whatever the secondary missions of their companies might say, these are for-profit businesses with shareholders and all that involves. As long as that is the case, we will have issues with it.

I encourage the Minister of State and I will give her my copy of a brilliant book by Bruce Schneier titled Rewiring Democracy, which really helps to unpick this. It is a brilliant piece of work by a cybersecurity expert. I have quoted it a few times as it is such a helpful book. It is a very positive take on it and it says how this is technology that can be good in society but it will only ever work if we trust it. Otherwise, we will see a complete divergence between people paying to have technology they can trust and the rest of us swimming around in the cesspit that the Internet is increasingly becoming.

Before I go into specifics of the Bill, I want to touch on the AI Act and the digital omnibus because I do not support the digital omnibus. It has moved what could have been more promising legislation back towards light-touch regulation and that is a problem. I appreciate that there were a couple of carve-outs at the last minute, which meant that the digital omnibus was not as negative as it could have been. Nonetheless, I think it has cut the AI Act at the knees before it has even started to get going. That is a real shame. Specifically on the Bill, I want to talk about some of the amendments I have submitted, which we will get a chance to discuss further on Committee Stage.

While it is obviously welcome that we are legislating for artificial intelligence, I note it is disappointing that such vital and important legislation is being rushed because of what I can only call inadequate advanced planning. The EU AI Act applies from August and we have had years of lead-in time to legislate for that application but yet this 106-page Bill was published late last week and the deadline for amendments for Committee Stage was this morning at 11 a.m. We have to get legislation like this right and to do that we need proper parliamentary scrutiny in the establishment of new State bodies and a complex framework of regulation. We need a process that will create space and time to identify key improvements for this Bill. That is extremely difficult to do when legislation is rushed through these Houses in the manner we are seeing right now.

Governance and independence of the AI office is something I have commented on a lot. I will continue to work on it throughout the passage of this Bill and it pertains to the bulk of the amendments I have put forward. I have spoken at length about the importance of the AI office as a regulator having the appropriate independence and resources to operate. That means it can act appropriately in its functions and that it is insulated from potential corporate capture, like we saw with the environment agency in the UK, or pressure from those it seeks to regulate. It needs independence from the Government like other regulators in this country. Its board and leadership must be appointed in a transparent manner and we cannot have a situation where such a vital piece of our regulatory framework - a framework that affects all parts of Irish society - is shaped wholly by the wants and interests of the Department of enterprise.

I was the head of an independent State agency. I understand what that independence looks like and how it can make bodies so much more effective. I raised this in committee but I also encourage the Minister of State's Department to look into the Paris Principles as essentially a gold standard in independence. The Irish Human Rights and Equality Commission is the only agency in the State that applies the Paris Principles and that is because it is a UN accredited body, but there is no reason we cannot apply those same standards to other bodies.

As it stands, this Bill suggests that the first board and CEO be appointed by the Minister for enterprise with little reference to any binding recommendations from the Public Appointments Service or procedures in the appointment to safeguard independence from the beginning. This is not for a caretaker or interim period before permanent appointments can be made; it is for the full five-year term. If we want independence, transparency and due regard for all sections of society that are impacted by the proposed remit of the AI office, we cannot have all that power and discretion over the people who will shape the direction and the governance of the office resting solely with the Department of, or the Minister for, enterprise.

I have concerns over that control within the Department because I have been reassured time and again by the Government that the AI office will be wholly independent. However, in my reading of this Bill I do not see how that can be the case because if we are serious about having an effective AI office and an effective administration of the AI Act, we need to address the gaps and deficiencies in this Bill and I hope we are able to discuss the constructively on Committee Stage.

The idea that the Minister can instruct the office to act as he sees fit, even with carve-outs for specific enforcement matters, undermines that independence. We cannot have an effective, independent State body that must provide clear and evidence-based views and guidance on the regulation of artificial intelligence that is not permitted to comment on Government policy when asked by Members of these Houses at Oireachtas committees. These are very basic and straightforward infringements on the potential independence of the AI office under this Bill. We have a myriad of examples of agencies and bodies in this State where these constraints are not applied and we see the benefit in ensuring accountability, boosting public confidence and giving us all clarity when it comes to legal and policy areas that need work. I want the perspective of those working in regulating AI to inform wider society, us included, but that is not possible if they are explicitly prohibited from sharing their perspective.

Next, I wish to speak to high-risk systems and ensuring that dangerous products are taken off the market. I welcome that we will eventually have a clear way of dealing with high-risk AI systems, but we need to be able to reflect reality. There are high-risk systems already in use that should not be there and waiting for legislation to catch up is just not okay. Under the digital omnibus plans that the Government has said it supports, we will be waiting another year until the regulations on these systems apply. That means we are essentially inviting companies to enter their products into an unregulated space within the next year before the regulation kicks in in 12 months' time when we will have rushed products and where developers' own safety processes will be rushed.

These are not simple, harmless products. They are explicitly the type of AI product which poses a high level of potential risk to our fundamental rights and safety. We, as a State, and particularly as a country where a lot of AI and tech companies are based, should be pushing for these safeguards for our citizens to be introduced and bedded in as soon as possible. With deepfakes, AI scams and more already affecting people, we cannot let these extremely powerful business interests have their way when it comes to the current status quo of no regulation, light-touch regulation and, in some cases, even self-regulation.

We also need to see clarity when it comes to the remit and co-ordination of market surveillance authorities and fundamental rights bodies because I understand and appreciate the reasons for adopting a distributed approach to the implementation of the EU AI Act but as others have mentioned, this approach is only effective when the national co-ordinating body, the AI office, can act as that co-ordinator effectively, that is, when it can mediate disputes between bodies where there are competing investigations and ensure smooth co-operation between the bodies involved. This is something that was flagged during pre-legislative scrutiny and which I am disappointed to see was not addressed in the Bill itself. I hope we can address this gap as the Bill proceeds through the Houses as it is absolutely vital to ensure that businesses, citizens and regulators have certainty on where they stand, how investigations proceed and who is in charge of those investigations in each case.

On that last point, it might be helpful for the Minister of State and her officials to have a look at the national preventive mechanism, NPM, structure. It is not actually legislated for here in Ireland yet. This is for the optional protocol to the convention against torture, OPCAT. It is actually quite a similar model where there is a co-ordinating body at the top and then that is co-ordinating sectoral inspectors. The legislation will eventually come in here in Ireland but there is a similar model in New Zealand that the Minister of State can examine. Basically it says that, because certain agencies will have greater powers in terms of independence, powers of publication and so on, they should then be able to co-ordinate bodies that do not have that same level of autonomy or independence. They may be further embedded into Departments, for example, but have quite specific roles to play because obviously, with the market surveillance authorities and the fundamental rights bodies identified within the Act, that is going to be the case.

As there will be a mixture of independence and autonomy within each of those agencies, making that co-ordinator as independent, autonomous and well resourced as it can be to do its job is absolutely crucial. Again, having been in a body where these functions are designated to you and assigned to you, rarely are they matched with the resources needed to do the job. We need to make sure that those fundamental rights bodies that are listed in Article 77 have the actual resources to do their job well. I have a range of amendments submitted and I look forward to discussing them on Committee Stage.

Deputy Naoise Ó Cearúil

Gabhaim buíochas leis an Aire Stáit as an mBille seo a thabhairt os comhair na Dála. I welcome this Bill and want to start by commending the Minister of State and her officials on the considerable work behind it. As a member of the Oireachtas Joint Committee on Artificial Intelligence, I have watched this framework take shape and I want to state clearly that its arrival is good news. It puts our implementation of the EU AI Act on a proper statutory footing and it does so at a moment when certainty matters.

The Act requires member states to have their market surveillance authorities, MSAs, in place by August 2025 and the high-risk rules apply from August this year. Businesses, public bodies and citizens have been waiting for this clarity and this Bill begins to deliver it. There is a great deal to welcome. The establishment of Oifig Intleachta Shaorga na hÉireann will give us, for the first time, a national home for AI co-ordination, literacy and public awareness. The regulatory sandbox and the real-world testing provisions in Part 3 are genuinely pro-innovation and they give smaller Irish companies and start-ups a supervised path to develop responsibly. That balance of safeguarding people while supporting innovation is exactly the balance we should strike.

I also welcome the careful enforcement process in Parts 5 and 6 with proper notice, the right to make representations and appeals to District Court and confirmation of fines by the High Court. This is firm but fair and it respects due process. I welcome that the Bill names our fundamental rights bodies in section 53 with the Data Protection Commissioner, IHREC, Coimisiún na Meán and An Coimisiún Toghcháin among them, with a clear route to test high-risk systems.

So, the foundations are strong and I support this Bill. I rise not to oppose it but to help strengthen it. In that collaborative spirit, there will be a need for some amendments, in my opinion, on Committee Stage and I want to set out some of those that I believe the Minister of State and her officials should be considering, and that we will consider, on Committee Stage with this House.

The first concerns the office itself. Under section 9, it is provided that Oifig Intleachta Shaorga na hÉireann co-ordinates, advises and promotes but it does not investigate, inspect or enforce.

Those powers rest with the designated sectoral authorities. I understand the logic of a distributed model and there is much to be said for drawing on the deep expertise of our existing regulators but there is a risk that the body the public will see as the national AI regulator will have no power of its own to act. There is a related point in section 42, which states, "The Office is designated as the single point of contact", yet Article 72 of the AI Act provides that the single point of contact is to be a market surveillance authority. I will bring an amendment to clarify the office's mandate in order that it can co-ordinate with real authority and our single point of contact will sit comfortably with Article 72. I would welcome the Minister of State's thoughts on that as well.

The second concerns co-ordination. Co-operation across several authorities and the Central Bank depends, under section 63, on voluntary co-operation agreements, and section 63(8) provides that a failure to comply with such an agreement does not invalidate the exercise of any power. Co-operation is welcome, but co-ordination that rests only on goodwill can become co-ordination on paper. Again, I will bring an amendment to provide a modest backstop so that where authorities overlap or a gap appears between them, there is a clear means to resolve it rather than hoping they agree it between themselves.

The third concern for resourcing is that Article 73 of the AI Act requires our competent authorities to have adequate technical, financial and human resources. The Bill provides for funding of the office but it is quieter on the capacity of the sectoral authorities that will carry out the enforcement on the ground. Again, I will bring an amendment for regular reports to this House on the resourcing of those designated authorities, which goes beyond the Minister of State's remit and, obviously, to the competent authorities.

I will add one further point. Section 9 provides that the office is independent while section 11 allows the Minister to direct it. I simply ask that, as the Bill progresses, we ensure such directions can never reach the operational or investigative judgment of the office.

I offer all of this in genuine partnership. This is a welcome and important Bill and I want to see it enacted and working well for the people we serve. I look forward to working with the Minister and my colleagues across the House and on the AI committee on Committee Stage, and to making the Bill even stronger.

Deputy James Geoghegan

I commend the Minister of State on bringing this Bill to these Houses. I want to make one point about the Oireachtas AI committee. Unfortunately, as the Minister of State knows, this was a Bill that did not go to the Oireachtas AI committee; it went to the Department of enterprise committee. This is not within the Minister of State's gift but I do not believe this is the last time in these Houses that we will be regulating or legislating in the area of artificial intelligence. I think the work we have done on the AI committee demonstrates a level of knowledge and understanding in relation to these issues. In the future, we should look at making it a permanent committee that would have the role and function of scrutinising legislation like this.

Unlike some other speakers, I like the aspect of the Bill whereby the Minister can direct certain aspects of how the AI office would operate outside of the regulatory and enforcement level. I think we should be more specific as to what kinds of directions we are going to give the AI office. Right now, we know what the AI office is going to do. It is going to enforce the EU AI regulation and make sure that enterprises are compliant in relation to it. What we do not know is the role it will have in regard to advocacy. I would like to see the Bill reflecting an AI office that would work on AI literacy, for example, and would have a clear, stated function in the whole area of AI literacy for both enterprises and consumers.

Ultimately, the fears and apprehensions people have about AI are similar in industry and to people just living their daily lives. There is a great fear about what AI is going to bring and yet people, particularly in Ireland, have taken up its usage quite rapidly. In the OECD, we are even making improvements in respect of our small business adaptation, yet that fear remains and prevails. Quite clearly, this new AI office, perhaps working with Enterprise Ireland, could have a very strong role in delivering a wide literacy programme. However, AI literacy is more than just enterprise. It is in our primary and secondary schools. What role will the AI office have in respect of AI literacy everywhere?

What we are doing at a regulatory level with the AI office makes sense. We are going to have a number of competent authorities that are going to have a regulatory function as it relates to AI within their own specialist domain and yet we are still going to have this single, overarching AI office when it comes to regulation. Surely we should have the exact same thing when it comes to advocacy, understanding and AI literacy that while it might not be the AI office delivering AI literacy schemes in schools, it could be the AI office that advises and directs the Department of education. It could be the AI office - if perhaps not meeting with small business to foster greater understanding of their adaptation of AI - playing a role in working with Enterprise Ireland to ensure that Enterprise Ireland is sufficiently tooled up to provide those necessary levels of adaptation if we are going to remain competitive.

I might conclude with the wider context. It is a good thing that we are introducing laws in the area of AI. It should help grow and foster trust among the public as regards what we are doing but let us be real here. There are massive fears about the impact of AI in regard to unemployment. I know SOLAS is establishing a national skills observatory under the Department of further and higher education but there are even greater global fears. They talk about superintelligent AI being a global systemic risk. Even industry is now kind of saying the same things as people were perhaps previously in industry and are critical at the pace of growth among these frontier AI companies. They are talking about a moment similar to the nuclear arms race, where the world needs to come together and agree certain levels of norms.

Right now, the European Union, with 700 million people, has agreed certain levels of norms but if the United States, China, India and other major global powers are not all in agreement as to how we are going to regulate AI moving forward, if we are at a certain level of superintelligent AI, God knows where we will be in 20 years. If we do not deal with that challenge, everyone will suffer. I hope that in our hosting of the EU Presidency, the AI summit, our enactment of the laws today and the establishment of the AI office, that we can play some role in that major global decision and that we can use the Presidency to do that.

Deputy Malcolm Byrne

I welcome this Bill and I thank the Minister of State and her officials for the work on it. I note that there was a fairly tight timeframe for amendments so I hope, on Committee Stage, there will be consideration of some of the issues here. While I commend the Minister of State on her understanding of the implications of AI right across society, I am concerned that there are certain Departments that have not yet got sufficiently up to speed and are not sufficiently invested in this area.

To speak specifically to the Bill, I welcome the fact that the AI office is, in a sense, positioning itself as a key hub and that it is going to be regarded not as a "thou shalt not" regulator but as an enabling body. I certainly welcome the distributed regulatory model that is proposed within this legislation. There is sector-specific expertise within all of our existing regulators and it needs to be supported to be able to ensure - for instance, if it is the Central Bank and it is dealing with a financial product but in these circumstances it is an AI-designed financial product - that it can regulate it in the same way as every other financial product. The legislation is quite silent on this and the Minister of State might respond on it, but that requires ensuring that those regulators and those who are identified within this legislation are adequately resourced both in terms of cash and, more importantly, personnel. That needs to be addressed.

In regard to some of the functions, it needs to be seen more as enabling. I welcome the provisions around the regulatory sandboxes but how this will enable businesses to safely use AI, certainly in respect of the educational role and the whole area of what I would call AI fluency, is essential.

There is one thing I would like to see in this. The Minister of State knows that when she worked with me in the context of the Online Safety and Media Regulation Act, we provided, when Coimisiún na Meán was being established, that there would be a children and young people's advisory panel. That was established. The Minister of State knows that it worked very successfully as well with the Ombudsman for Children's Office. I think it is critical within this legislation. It should be put on a statutory basis that we have a children and young people's advisory panel and that they are very specifically identified as a vulnerable group but also people who are going to be very engaged with this technology. I ask that on Committee Stage there would be statutory provision for such a panel in the same way as we have seen with some other bodies.

I welcome the definitions in respect of the chief executive. I understand recruitment for the role is under way at the moment. On the appointment of whomever that individual will be, he or she will, in many ways, set the direction for this body. I appreciate the Minister of State will have certain roles in being able to direct how the office is going. The message that has to go out there is about how Ireland will balance support for innovation on the one hand with safe regulation on the other, and that this office will be an enabling one rather than an office that will seek to shut down the possibilities that AI provides.

Deputy Martin Kenny

The whole world has changed so much in the past decade. Certainly, the advent of AI has been a huge advance, in one sense, in respect of all of that. I recently spoke to somebody who has developed a business in the United States in regard to the whole issue of cybersecurity and AI is central to what they do in respect of all of that. We also, of course, know that it has the ability to do repetitive tasks without ever getting fatigued or tired or lazy. It can do all those things. It can assess vast amounts of data and zone into what is effective. It has huge potential and that has to be acknowledged as well. It is not a force for good only; it can also be a force for great harm. It depends on how it is used and who uses it. That is where human intent comes into play in respect of that.

That is why it is important that we have regulation. We often talk about the enforcement of regulation and enforcement to ensure that the negative is not enabled in respect of this but is disabled. It also has to have an element where the enforcement of that regulation has to be able to ensure that the positive is enabled and that there is a means by which people can easily access and bring forward ideas and prospects and that the means of using artificial intelligence are for the good and advancement of people. We must ensure we have better systems in place that can deliver for populations.

The big issue people often come up against is that it can have a very negative effect in that it will take away jobs and all of those things. That is always, of course, an aspect to it. However, there are so many of the tasks we cannot find people to do that artificial intelligence will never be able to do. I refer to healthcare and a whole range of areas where we absolutely will never be able to do any of those things without having human beings to do them who can show care and compassion and all those things. Artificial intelligence does not have that. That is what we need to acknowledge and understand.

While the whole issue and one of the big focuses in respect of this is safety and security, it is also about ensuring that the negative aspects of it are not allowed to overtake to become the dominant feature of it. So much of what we see and understand of the world of cyber and the Internet has become so negative. It has become so damaging and so evil, one could say, when we see some of the stuff that is going on there. We want to ensure that artificial intelligence is prevented from - it is already starting to do it - magnifying and exaggerating and pushing of those negative things even further. That is a big aspect of what to put in place here. Of course, there also has to be accountability, fairness and integrity around it. Having an office in place and having laws and regulations in place do not provide any of those things unless we ensure that we have the intent and we are absolutely sure we have the ability. That is about staffing. That is about funding. That is about making sure we can have the regulations but that those regulations are applied rigidly, particularly in situations where we have large groups in different parts of the world that can hide behind each other. We talk about the cloud. Where is the cloud? Where does it exist? How does one get access to it? There are big issues here that need to be dealt with. Artificial intelligence can actually be used to deal with those difficulties and that is where it needs to be done positively. The concept of this is welcome, but we have a distance to go. I am particularly concerned with regard to the funding and staffing of the offices. I am concerned that we may not have enough time to get into this race because we are in a race against very negative actors who are continually trying to undermine the very aspect of civilisation. They are trying to set us back into the past and we need to make sure that does not happen.

Deputy George Lawlor

We in the Labour Party welcome the intent of this Bill. We are far from the first to say it, as the Minister of State knows, but the emergence of artificial intelligence could be the most significant societal and economic change we will see in our lifetimes. AI is already reshaping how decisions are made that affect people's lives - decisions about credit, employment, healthcare and policing. It is reshaping our economy, our jobs market, the way we work and, fundamentally, how we live our lives. The pace of change is only set to pick up. Sadly, we see AI almost on a daily basis attacking jobs, particularly in the tech sector. Governments around the world need to recognise the scale of the change and challenge here and respond. It is not enough to leave this transformative change in the hands of the tech giants whose sole priority will be the power and profits they hope to reap from this transformation. Regulation is essential. Democratic guard rails are essential. Governments are facing a choice, not between AI or no AI but between AI that serves the people or AI that serves the tech billionaires. The false dichotomy that some powerful interests would paint between regulation of AI, on one hand, and innovation on the other needs to be rejected. AI has the potential to deliver so much good in healthcare and how we deliver services and productivity but to unlock any of those benefits requires public trust. Securing that public trust requires effective regulation.

That is why the EU AI Act was such an important world first in beginning to put regulatory guard rails on how we use and develop artificial intelligence. It was pushed for by socialists and democrats in Europe who recognise that in the face of a new industrial revolution, we need to put the safety and well-being of citizens first. We need to keep humanity at the centre of AI policy. The EU AI Act is not perfect. There are issues around companies effectively self-designating as high-risk or otherwise and around how it fails to effectively safeguard copyright protections for creatives, to name but two. There are gaps in its definition of high-risk AI uses, such as chatbots used by children. It was, nonetheless, a hugely important step for Europe to take.

The Bill aims to implement some of the requirements on member states arising from that legislation as they relate to Ireland. That indeed is a welcome aim, and the establishment of Oifig Intleachta Shaorga na hÉireann is to be welcomed. However, we need to make sure this legislation is watertight. The regulation we put in place needs to be effective and workable.

It is in that light that while I support the ambition of the Bill, I want to raise a fundamental concern about one specific part of it. The Bill creates a system of very large financial penalties, fines that in some cases could run to tens of millions of euro or a percentage of a company’s global turnover. It provides that these fines are to be determined not by a court, but by a person called an adjudicator who is appointed under the Bill and who operates within the regulatory system. The role of the courts comes late. By the time the courts are involved, the critical decisions have already been made. In Ireland, we have had systems of administrative fines in financial regulation, competition law and data protection for many years. This Bill pushes the model further than it has been pushed before. At some point, the model breaks, not as a matter of legal technicality but as a matter of basic constitutional principle. As the Minister of State knows, our Constitution is built on a separation of powers. The judicial function is reserved to the courts by Article 34 of the Constitution. It cannot be handed over to a civil servant or regulator or an adjudicator, however well-qualified and however independently they act. The reason is that the independence of the courts and the procedural protections of court proceedings are structural guarantees that protect the citizen from the power of the State. They are not formalities.

With that background in mind, let me set out the four concerns we have. The first concerns the nature of the penalties themselves. Our courts recognise that not everything attracting a financial penalty is a criminal matter in the constitutional sense. A fine imposed by Revenue for late filing of a tax return is not a criminal conviction. A penalty imposed by the Central Bank on a financial institution for a regulatory breach is not the same as a criminal prosecution. The courts have drawn a distinction, in broad terms, between penalties that are genuinely regulatory in character, aimed at securing compliance with a scheme that the regulated party has voluntarily entered and penalties that are, in substance, punitive, aimed at identifying, denouncing and punishing wrongdoing. These punishment cases require the full protection of a criminal trial before an independent court.

The difficulty with this Bill is that the fines it contemplates do not sit comfortably in the regulatory category. They can be absolutely enormous. The Al Act provides for fines of up to €35 million, or 7% of a company’s total worldwide annual turnover, for the most serious of violations. For a large technology company, 7% of global turnover could be billions of euro. No court in Ireland has ever had to decide whether a penalty of that magnitude can be described as regulatory rather than punitive. I am advised that when looking at a fine that could put a company out of business entirely, the word regulatory starts to lose its meaning and the constitutional protections that attach to criminal proceedings start to apply.

There is also a difficulty with the consent argument that sometimes justifies administrative penalties in other regulated sectors. For banking and financial services, there is at least an argument that companies entering a licensed sector applied for a licence and knew what they were signing up for. That argument simply does not work here. The Al Act applies to anyone who develops, deploys or uses certain types of Al system. There is no licence to apply for. The consent justification, which has some logic in the banking context, is not available here.

I should also address directly the argument that because this Bill gives domestic effect to an EU regulation, our domestic constitutional concerns do not apply. That argument does not work in this case. The EU Al Act requires lreland to have a penalty regime, but it does not prescribe how that regime must be structured. The choice to use an adjudicator model with a court confirmation mechanism is an Irish choice, made by the Irish Government, and it falls to be assessed against the Irish Constitution. We cannot borrow the authority of EU law to excuse a domestic structural choice that EU law did not require us to make.

The second concern goes deeper and it does not depend on the penalties being characterised as criminal. Even if one accepted that these are purely administrative rather than criminal penalties, the process of imposing them may still amount to what the Constitution calls an administration of justice, and the Constitution permits only limited functions of a judicial nature to be exercised by non-judicial bodies. Our courts addressed this question directly in a case called Re Haughey in 1971. That case arose from proceedings before a Dáil committee. The Supreme Court held that the label attached to a proceeding does not determine its character. What matters is the substance. Is a body making findings of fact about a named individual, with legal consequences adverse to that individual? If so, the proceeding has the character of an administration of justice. Then, the constitutional rules that govern the administration of justice apply regardless of whether the body is called a court, a committee, a tribunal, or, I would add, an adjudicator.

Applied to this Bill, the argument is straightforward. An adjudicator under Part 6 makes findings of fact on whether the company breached the Al Act and, on the basis of those findings, imposes financial consequences of enormous severity on a named legal person. That is, in substance, an administration of justice. The adjudicator is not merely advising, not merely investigating or not merely recommending; the adjudicator is determining. I am advised that a body that determines justiciable controversies with serious consequences for individual rights is administering justice, whatever it is called. Again, by looking to the size of the available penalties, it is very hard to see how this could be classed as a limited exercise of a judicial function.

The third concern arises directly from what the Supreme Court said in Re Haughey about the mechanism used in that case to involve the courts. The Dáil committee in that case could certify a person for contempt to the High Court, which then had jurisdiction to punish. The argument was made that this saved the constitutional position. The committee investigated and certified that the court punished, therefore. the judicial function was exercised by the court after all. The Supreme Court rejected that argument, however. It said that what was being sent to the High Court was not a case to be tried. The trial, such as it was, had already taken place. What was being sent was effectively a person to be sentenced, and a sentencing hearing before a court that has not itself tried the matter is not an administration of justice. The court was being used as a rubber stamp, not as a decision maker.

We can compare that with the confirmation mechanism in this Bill. Under sections 110 to 114, when a fine imposed by an adjudicator is not appealed, the market surveillance authority must apply to the High Court to confirm it. The High Court must confirm unless there is a fundamental and manifest error of law or the fine is clearly excessive or disproportionate. The evidence available to the High Court is limited to what is before the adjudicator and the adjudicator’s findings are binding. The High Court does not rehear the case and it does not make its own findings of fact. It reviews, on a deferential standard, a determination already made by someone else. The parallel with Re Haughey is exact. The adjudicator tries and the High Court merely confirms. The findings of fact are not revisited. The evidence is not re-examined. That is a sentencing hearing, not a trial. I am also advised that it is open to precisely the same constitutional objection that the Supreme Court upheld in Re Haughey, in that the court is being given a jurisdiction to endorse a determination made elsewhere and not a jurisdiction to determine anything itself. If the process before the adjudicator is constitutionally infirm for that reason, nothing the High Court does at the confirmation stage can cure it. A flawed process does not become constitutionally valid because a court subsequently endorses it.

The fourth concern is that constitutionally it is for the court, and the court alone, to make findings of fact in proceedings with serious consequences. A statute that requires a court to accept as established a fact found by someone who is not a judge, in proceedings that were not a trial, is removing the fact-finding function from the court and that, the Supreme Court said, is unconstitutional. I am advised that the confirmation mechanism in this Bill is open to precisely this objection. The High Court at the confirmation stage is bound to accept the adjudicator’s findings of fact. It cannot go behind them. It cannot hear fresh evidence that contradicts them. It must treat those findings as established and proceed to confirm the fine on that basis, unless the error is fundamental and manifest. The facts have been found elsewhere by a non-judicial body and the court is required to accept them. The High Court cannot be given a jurisdiction that presupposes the facts have already been found.

Taken together, these four concerns point in the same direction. The confirmation mechanism is not a genuine exercise of the judicial function. It is a review of a determination made elsewhere, on a deferential standard with a closed evidential record, in respect of findings of fact that the court is not permitted to revisit. That is not what the Constitution means by the administration of justice. I am also advised that a court exercising that kind of jurisdiction is not curing whatever constitutional deficiency existed in the adjudication process. It is merely adding a judicial gloss to a process that remains, in substance, non-judicial.

I invite the Minister of State in her response to engage with the constitutional question directly. These are not peripheral concerns. They go to the central mechanism by which this Bill proposes to enforce the Al Act in Ireland. I appreciate the Bill has been carefully prepared, but careful preparation and constitutional validity are not always the same thing, and on this particular point the Bill, as drafted, is on genuinely uncertain ground. We absolutely need to get this right. Al, as we all know, is rapidly changing the world around us. Regulation is vital to ensure that ordinary people are protected and proper guardrails are in place. That idea is not in conflict with innovation or adoption; it is essential to it. Europe has led the way globally on this. When we are now looking to implement those regulatory guardrails in our own State, we need to be certain that we are on firm ground. It is too important to get wrong. As I said at the outset, we in the Labour Party welcome the intent of the Bill, but we simply need to get it absolutely watertight and right.

It is far too important not to. I look forward to the Minister of State's response. I commend the principles of the Bill to the House.

Deputy Erin McGreehan

I very much welcome the opportunity to speak on the Regulation of Artificial Intelligence Bill. I commend the Minister of State on bringing forward such important legislation and congratulate her on all the work she does in her brief. I wish her continued success in this area.

As we all know, artificial intelligence is transforming how we live, work, learn and access services. It offers enormous opportunities in healthcare, education, scientific research, agriculture, and business innovation and productivity. However, those opportunities come with significant responsibilities. This Bill provides a necessary framework to implement and enforce the EU Artificial Intelligence Act in Ireland. It establishes the AI office in Ireland as an independent statutory body and ensures that appropriate oversight, accountability and enforcement mechanisms are in place. Importantly, it does not create additional burdens beyond those already provided for under EU law. Rather, it ensures that Ireland can meet its obligations effectively.

As Chair of the Oireachtas Joint Committee on Further and Higher Education, Research, Innovation and Science, I particularly welcome the establishment of the AI office. I believe its role should be extended beyond regulation alone. The office should become a national co-ordination hub, working closely with organisations like Skillnet Ireland, Enterprise Ireland, our local enterprise offices, higher education institutions, further education and training providers, schools and industry partners to drive AI awareness, adoption and skills development across our entire society.

The reality is that AI is not something for the future; it is already here. Our challenge is to ensure that individuals, businesses, public services and communities are prepared. We need to build AI literacy across the population so that people understand both the opportunities and all the risks. At the same time, we must ensure that workers, students and enterprises have the practical skills needed to harness the benefits of these technologies. If we get this right, AI could help us improve public services, support innovation, increase productivity and deliver breakthroughs in research, healthcare and education. However, those benefits will only be realised if we invest in people and equip them with the confidence and skills to engage with AI effectively.

We must ensure transparency, accountability and inclusion remain at the heart of all of our approaches. New technology must empower people and not exclude them. That means paying particular attention to accessibility, disability inclusion and equal opportunities as AI systems become embedded in everyday life. AI must always serve the public good. Innovation cannot come at the expense of public trust, child safety or the rights of vulnerable people. Technology companies must be properly regulated and high-risk AI systems carefully monitored. We must also ensure that workers whose jobs may be affected by AI are supported to retrain and adapt, including through greater use of the National Training Fund to provide flexible upskilling and reskilling opportunities.

Ireland has a huge opportunity. I am hugely ambitious about our role in this. We can be a leader not only in AI innovation, but also in AI readiness. This legislation is a really important step in building the foundations for that future. I am very pleased to support it.

Deputy John Clendennen

I welcome this Bill. It underscores Ireland's commitment, under EU direction, to a future where technological advancement will be inextricably linked to ethical governance and societal well-being. The commitment to implementing this through the EU AI Act is about protecting citizens and building trust around AI. More often than not, we hear too much doom, gloom and negativity around AI. We hear about the threats it poses rather than the opportunities it presents. We must focus on ensuring the regulations are robust in that regard. The context makes this Bill not a mere legislative formality, but our collective national response to artificial intelligence as we ensure that we implement the EU AI Act effectively and responsibly. I want to focus particularly on responsibility this evening.

We have discussed SMEs at length in the enterprise committee. Let us look at some statistics. SMEs are the backbone of the Irish economy, yet evidence shows a growing divide in AI adoption between larger companies and smaller firms. Recent data from the CSO shows that while 58% of large enterprises are using AI, only 17% of small enterprises are doing so. Ireland currently ranks only mid-table in Europe for business AI adoption. If we are serious about competitiveness, we must close that gap.

Should the AI office have a stronger statutory mandate than is currently provided for in the Bill? At present, the legislation contains extensive provisions regarding investigations, sanctions and enforcement. By contrast, there is no real focus on supporting businesses to adopt AI responsibly and successfully. The AI office should have a stronger enterprise support mandate. This would include providing practical guidance to businesses on compliance, delivering training and awareness programmes, advising Government and State agencies on measures to increase SME adoption, and promoting innovation, competitiveness and responsible deployment. We must be cognisant of the fact that regulation and innovation are not competing objectives. If implemented appropriately, they complement each other.

The Data Protection Commission could also have a greater role. It has protected citizens and their rights through the enforcement of the GDPR. However, as AI becomes more deeply integrated into business operations, organisations increasingly need practical guidelines and guidance on how to comply with both data protection requirements and emerging AI rules. Strengthening the DPC's mandate to include a clearer business support and guidance function would complement rather than diminish its independence as a regulator. I encourage the Minister of State to look at the amendments to strengthen the enterprise support mandates of the proposed AI office and the DPC.

I acknowledge our position globally. We will ensure that protections in high-risk areas such as biometrics, education and employment are robustly implemented by the December 2027 deadline. Considering our location and that Ireland is home to eight of the leading foundational AI model providers globally, we are uniquely positioned to be a hub for AI innovation. Implementing a functional, clear and effective regulatory framework will not only meet our EU obligations, but also strengthen our position as a leader in responsible AI development, which will, in turn, attract further investment and foster economic growth.

This Bill demonstrates our commitment to online safety and the promotion of trustworthy AI innovation. I commend the Minister of State on bringing it to the House. I very much welcome it.

An Leas-Cheann Comhairle

Before I call on our next speaker, I welcome all of those in the Public Gallery. I hope they are enjoying the debate and finding it informative. They are very welcome to Leinster House. I particularly welcome Peter and Briege Rust from Dundalk, who are here with Deputy McGreehan, and Deputy Cormac Devlin's friends and neighbours.

Deputy Cormac Devlin

I thank the Leas-Cheann Comhairle for that. I welcome the opportunity to contribute to the debate on the Regulation of Artificial Intelligence Bill 2026. As the Taoiseach noted, the accelerating growth of AI is both a challenge and an opportunity. AI is here, being used daily by businesses and individuals, and it will continue to develop. It is not something we can ignore. AI-related economic activity is already beginning to materially influence the Irish economy.

I support the Government’s ambition to make Ireland a global hub for applied AI innovation. The Bill implements European legislation in this area, establishing the AI office of Ireland, which will co-ordinate efforts to supervise the sector. In itself, legislation is insufficient. We learned that from the failures relating to the banking sector and social media policy. Given Ireland’s position as a base for many companies, it is important that we support the sector, but it is also critical that regulators take an active role, proactively working alongside major providers to ensure algorithms and products benefit the public good and do not create undue risk.

More broadly, I will raise the issue of access. Together, the EU and UK comprise a market of almost 520 million people with a combined GDP of almost €24 trillion. We need to see a far more robust response from the European Commission to ensure businesses and individuals have access to the latest software to remain competitive. I hope Ireland will take this up as a priority for the upcoming Presidency. Europe cannot be allowed to fall behind or be locked out of essential advanced products and services. Together, the US, EU and UK comprise the largest global market by far. It is in everyone’s interest that we work together, remain competitive and deal with the challenges and opportunities presented by the growth of AI.

Deputy Paul Donnelly

I welcome the Bill and the opportunity to speak on it. It is critically important. AI is something we use every single day, whether we know it or not. Many of the applications we use have a major component of AI. In many areas it has a really useful purpose. Students and workers use it, and people in here use it on a daily basis. As I said, AI has a really useful purpose. I have watched a number of programmes on the use of AI in medical and pharmaceutical applications. This technology is driving real innovation and speeding up the time it takes to complete trials looking at different ways pharmaceuticals and medicine can help people. In that regard, it is really important and useful.

The genie is out of the bottle. AI is with us and it is not going anywhere. What we do with it is the most important issue. What this House does in terms of regulation and ensuring the companies driving all of this do so in the proper way and are monitored, with proper due diligence done on how AI impacts society, is very important. We know AI has positive impacts but it also has extremely negative impacts. We are talking about social media and how it impacts on young people, with different countries moving towards banning social media for young people. Sometimes we look at all of this in very simplistic ways. AI is deeply complex and something we are going to be learning about as it changes exponentially. It is moving at such a fast pace that what we talk about and think about today in relation to AI could be obsolete in a couple of months' time. That is how fast this is moving.

We need to make sure we have properly resourced structures in place. We are dealing with multibillion dollar companies which have vast resources and profits on a scale we have never seen. They have the capacity to tie up organisations such as the one we are discussing in legal issues. If we do not have the structures we are talking about in place and properly resourced and supported, our approach will not be effective. We will then be relying on the companies themselves to police themselves. That is the one thing we cannot do because we know they are driven exclusively by profit.

In relation to workers' rights, we are already seeing workers directly affected by AI. We need to make sure workers and young people are protected. We have talked about AI and social media use and the damage caused by algorithms and the lack of attention social media companies are giving to dealing with the harmful issues in relation to them. We need to make sure this body is regulated, well resourced and capable of taking on the big companies because we cannot trust them. That is the most important thing. We cannot trust these companies. They are driven by profit only.

An Cathaoirleach Gníomhach (Deputy John McGuinness)

I extend a welcome to Deputy Heneghan's former teachers and class who are in the public Gallery. They have a lot to answer for. They are very welcome and I hope they enjoy the debate.

Deputy Maeve O'Connell

This legislation seeks to ensure that AI transformation is guided by key principles, namely, safety, accountability and trust. We must also ensure that we extend those principles to the transformation AI is having on higher education. AI can, of course, supporting learning but it can also produce essays and other assessments. Now we have AI tools to help us detect AI-generated assessments. This has resulted in a weird arms race, with increasingly sophisticated AI tools to generate assignments and increasingly sophisticated AI tools to try to detect its use. Meanwhile, the impact of AI is threatening the success in widening access to higher education because in many cases, information technology is now being removed from assessments entirely.

The only comparator I have to the impact AI is having on higher education is Covid, which resulted in everybody having to work from home and teaching going online. However, there is one key difference. With Covid, there was always the expectation that eventually we would return to campuses and more familiar operations. It is not the same with AI. There is no return to the status quo there. This transformation is permanent.

The higher education system is not designed to withstand this level of a systematic shock. It is designed for gradual change and structured review. Assessment methods are regularly evaluated but these are informed by industry, research and engagement with stakeholders before any changes are implemented. This protects academic integrity and ensures students are taught what they need to prepare them for their futures. It is a core element of the design of system. This is not a system designed for overnight transformation.

Covid prompted a rapid shift to promote teaching and assessment and staff very much rose to the challenge. They transformed how they worked to support the students they had. However, it required an enormous amount of staff effort and commitment, which were never really fully recognised. It was the same for many other people who had to change how they did things during that time. After only a few years, we are asking those exact same staff to once again transform how they work.

Another key difference is that this time we have no established play book. We have no experts out there who have already solved all of these problems. Research is still emerging and best practice is still evolving. Meanwhile, our lecturers and researchers are on the front lines, addressing these challenges in real time on a daily basis. They are redesigning curriculum, rethinking assessment methods, responding to the misuse of AI tools and helping their students navigate AI ethically, all the while maintaining high standards of teaching and research. They are expected to do this in a system that was never designed to be this agile. This is an unrecognised and unacknowledged extra workload and responsibility on our academic staff. This adds to the existing workforce issues, where we have one in five academic staff employed in temporary or short-term contracts. We are now expecting those staff to navigate this changing world of AI and engage in all that adaptation, not knowing if they will have a job next semester or year.

If we are serious about ensuring AI is well governed, we need to invest in a national approach to AI in education because there are no ivory towers in a post-AI higher education world. We owe that to our third level staff and students and to the future of our higher education system.

Deputy Shane Moynihan

Tá áthas orm bheith in ann labhairt ar an mBille seo trathnóna. Ba mhaith liom moladh a thabhairt don Aire Stáit, an Teachta Smyth, agus an Teachta Ó Cearúil as ucht a gcuid oibre leis an mBille seo a chur le chéile.

I welcome the establishment of oifig intleachta shaorga na hÉireann, which will be a very important piece of the architecture to ensure the overall enforcement and monitoring of this legislation. My plea at this stage, as we establish that office, is that we staff it appropriately, get the appropriate talent and hire the appropriate professionals with the skills necessary to ensure we have strong and competent regulation in terms of how we engage with AI in Ireland. I emphasise that is not necessarily just saying we need to regulate. It positions Ireland as a very strong centre for companies and research centres that want to learn more about how to use AI. If we have the people and professionals in our regulatory and monitoring authorities who are able to watch, monitor and collaborate with that innovation, it will set us up to be a better centre for regulation, both in Ireland and across the European Union. Alignment with the European Union is extremely important in this instance.

The second point I make is with regard to the need to encourage innovation. LinkedIn data shows that Ireland is exceptionally attractive to AI engineers from all over the world. We have seen net positive migration of those individuals to this country over the past year.

That shows we have the skill sets and competence in this country and in that sense the oifig should be thinking about innovation. I encourage the Minister to consider whether there is scope to include in the Bill provisions to make oifig intleachta shaorga na hÉireann take the mandating and encouraging of innovation very seriously, not only in encouraging innovation around AI but also ensuring the regulation is fit for purpose. Often the criticism is made in this House and other places that regulation does not catch up with technology as it is developed. One solution would be to have the regulator sitting beside centres of innovation so it understands how that innovation takes place and how the resulting regulation can take account of that.

My third point relates to the need to ensure the oifig takes training and proficiency in AI skills as a core part of its mandate and in the roll-out of its work. All of the research on AI points to the fact that while AI technical skills are extremely important, human-centric skills, such as problem solving, team building and critical thinking, are also important. They are what makes AI such a powerful tool and it is important the oifig does not lose sight of that. I encourage those three points to be taken into account as the Bill proceeds to Committee Stage.

Deputy Ruairí Ó Murchú

I am delighted to have this opportunity to speak on the Regulation of Artificial Intelligence Bill 2026. It has been said by many previous speakers that AI is just a reality we are dealing with. Even though some people worry about it, we are a long way removed from Skynet coming live, if anyone remembers the glorious days of "Terminator".

We all know the issues with the tech sector. When we talk about AI, we know about the part that crosses over into the issues we are having with social media. We know there is a requirement for regulation. It is obviously vital we have oifig intleachta shaorga na hÉireann, the AI office of Ireland. We need a body to co-ordinate this. I get that in some level it will almost work against itself, in the sense that the first part of its work will be regulation, which will be required given all the issues we have seen with the tech sector lately and the fact we are not even entirely sure where AI is at now, never mind where it will be in two weeks' or two years' time.

It is very easy to talk about ensuring we promote AI literacy and have the proper regulatory sandboxes - controlled and contained situations where we can actually try to test AI. Obviously there will be a huge amount of international co-operation. We know who the big players are, namely, America and China. We know there is a certain element of open source and we are not entirely sure who wins out here. We know the huge dangers there are. The very simplistic dangers are the ones we know only too well. People will talk about Covalen and the loss of upwards of 720 jobs there. I have said it for a long time in relation to some of the jobs I have seen being lost in the likes of PayPal. I get the point that can be an issue of machine learning meeting AI. We have to have a strategy to ensure we do not miss out on the benefits of AI but we also have to deal with the displacement of jobs and other risks.

I attended a Children's Rights Alliance event earlier. A major part of the discussion related to where AI meets recommender algorithms in social media. We can all work on the basis that social media and big tech companies are about keeping you online. They will show you whatever they need to show you to keep you online. It is about you staying online and them making money. It is as simple as that. We know we will have to see a hell of a lot more from the Government, but also from Coimisiún na Meán, the European Commission and the Data Protection Commissioner, if we are going to see those companies brought to book properly. That needs to happen because we know of the huge dangers there are for children.

It was pointed out to me that, in dealing with this legislation, we have to make sure there is regulation of AI and protection and that we put in specific protections for children. The social media issue is about regulation. As others have said, as much as we want to see AI literacy, we also need to see digital literacy. We also need to see that people show a capacity and ability to critically analyse circumstances and situations. We need to ensure we teach-----

An Leas-Cheann Comhairle

The time is up, Deputy.

Deputy Ruairí Ó Murchú

I could go on for a lot longer but I will not.

An Leas-Cheann Comhairle

I understand Deputies Michael Collins, O'Flynn and O'Donoghue are sharing time equally.

Deputy Michael Collins

I do not think I will take the full amount of time I have. As a young fella growing up, AI had a totally different meaning in west Cork from what it means today. When we talked about AI, we were looking after animals and trying to bring a new type of animal into the world, but now AI refers to artificial intelligence. I do not know what they are going to call the artificial insemination man now. He was called the bull man down my way but they will have to call it some other kind of intelligence.

There is another aspect of artificial intelligence that we need to speak about plainly, that is, its misuse in everyday life. AI is a powerful tool when it is used correctly, or so I am told because I do not use it at all myself. It can help businesses to grow, improve services and make life easier for people. No one in this House disputes that. However, when it is misused we start to see real problems and ordinary people are the ones who pay the price. I am increasingly hearing from people who are on the receiving end of AI-generated communications, messages and letters that look like they have come from solicitors, are dressed up in legal language and are designed to put the fear of God into them. These are automated, often aggressive in tone and sent out in bulk. That is not professionalism; it is intimation by automation, and I have no time for it.

We are also seeing AI being used to waste people's time with endless, low-quality content, spam-type material and automated responses that add nothing valuable but create confusion and frustration. People are busy enough without having to deal with that kind of nonsense.

Then there is this issue of fake images and misleading content. AI can now generate highly convincing pictures that are simply not real. These are being used alongside exaggerated or completely false headlines. This is clickbait designed to grab attention rather than tell the truth. It undermines public trust, not just in technology but in information itself. Again, let me be very clear that I have no time for that either.

If we are serious about regulating AI, we cannot just focus on high level systems or large corporations. We must also recognise the everyday misuse that affects ordinary people, including the farmer, the small business owner and the pensioner who receives a threatening letter or sees a misleading image online. Regulation must send a clear message that AI is not a free licence to mislead, intimidate or waste people's time. There must be accountability when AI is abused. While we support innovation, we must also draw firm lines. AI should serve people, not frustrate, frighten or deceive them. That balance is what we need to get right. It is right, therefore, that we regulate AI. Independent Ireland accepts the need for clear rules, strong consumer protections and accountability.

However, supporting regulation does not mean giving the Government a blank cheque. The Bill creates a new State body with significant structures. The public deserves clarity on the cost and on whether we are duplicating existing regulators. We must also be honest about the burden on small Irish businesses. Large multinationals can absorb complex compliance rules. Local businesses in Cork and elsewhere cannot. We cannot create a system that suits big corporations while leaving Irish enterprises behind. The enforcement powers in this BiIl are substantial, including access to data and even source code. These must be exercised in a balanced and proportionate way to maintain industry confidence.

There are positives. The regulatory sandbox, AI literacy and the complaints system are all welcome but the key is balance. Ireland should be the best place in Europe for responsible AI innovation, protecting people while supporting enterprise. We support the objective of the Bill but we will press the Government on cost, proportionality and fairness for small businesses.

Deputy Ken O'Flynn

My colleague has spoken about the positives and negatives of AI. I want to concentrate more on the technicalities of the Bill.

I want to be clear from the outset what this Bill is and what it is not. Whatever we are being told it is certainly not a modest piece of housekeeping. Indeed, it is a scaffolding for an entire, new, State-run, artificial intelligence regulator, a Bill which now has 139 sections and four schedules, a statutory body and a sanctions regime reaching €35 million, or up to 7% of a company’s global turnover. That is what we are discussing here. Yet, this House is being invited to waive through this legislation on the understanding that it is just a little bit of housekeeping. That is the Government’s own framing of this. We are told that these are technically implementable measures. We are told that the real law is going to be written in Brussels, and that we are just there simply to be the building mechanism and the machinery behind that, to enforce something that will be taken from Brussels legislation and which has not yet been fully written. For that reason, and for that reason alone, to rush this Bill through should tell this House that for the reasons I have outlined we need to scrutinise this legislation far more closely, have far more time on this debate, and look at the legislation and the Bill far more than we are.

When a Parliament is reduced to installing the plumbing for a law that it cannot amend, the least we can do in this House is make sense of that plumbing work. Here in this Bill it does not. The Bill requires a new AI office of Ireland to be operational by August of this year. The State has set up an entire institution, it has to staff it, it has to fund it, and it has to switch the lights on within weeks. Yet, at a European Union level, the very obligations this office are meant to enforce are being pushed back under the digital ombudsman agreement, which I do not believe has been mentioned in this House, reached in May and not yet formally adopted, but already agreed between the institutions. The high-risk rules at the heart of this regime to be set up and applied will not be this August but in December 2027. We are here in this House and we are actually building a referee, at speed and with cost, 16 months before the match even begins. I put it to the Minister of State, Deputy Smyth, that we are legislating in the fog. That is what we are doing in this House at the moment with this information and this Bill. We cannot be told in this House the certainties and rules by which this office will finally have something to enforce. It is not coming in from Europe yet. We have not yet received that information.

Let us turn now to what is at stake for this country. I do not think we have played this out enough or thought enough about it. Ireland is not a bystander when it comes to artificial intelligence. We host it in American technology companies that lead the field and run their European Union operations from all across this country. They did not do this accidentally. The point, which the Government will not say out loud, is that last year corporation tax brought in the sum of €33 billion, which is a record amount. The top ten companies, overwhelmingly American multinationals, paid well over 50% of it. Around 75% of all corporation tax comes through United States firms. Those companies employ directly almost 200,000 people in the State, and another 160,000 indirectly in this State. That is not competitive advantage, or should be taken as casual trade. That is the single most concentrated precarious revenue stream in this State, and it must be protected and minded. With that in mind, what are we doing with it? We are volunteering to be the most diligent enforcer of a framework from the European Union that is imposed on an industry that it has failed to build itself. Europe has not built AI: America and China have. Europe is way behind when it comes to AI and AI firms. What are we doing here with the very firms we depend on for finance? Is the Minister of State aware that the independent Irish Fiscal Advisory Council has already said that tighter regulation in the sector will be a risk to the revenue receipts of the State? That is a very important statement and something we all need to reflect on in this Chamber. The Minister of State has met with these companies. They are mobile companies. They make location decisions every day. We should be giving them every reason to stay in this country rather than giving them reasons to exit this country. Instead, what we have decided to do through this Bill is implement and hand them a €35 million fine regime and 15 separate regulators to answer to. It seems bizarre to me to have 15 separate regulators to answer to.

Let me be straight with this House. I am not against oversight of technology in any way. No serious person in this House should be against oversight of technology, but artificial intelligence raises real questions and the public are entitled to real answers. A Parliament that truly cares about its citizens does not switch on an enforcement mechanism or an enforcement machine without knowing the rules around it first. It does not impose a cost on its taxpayers by putting at risk the very receipts that fund the Exchequer, that fund our social welfare, that fund our schools, that fund our third level colleges, that fund our hospitals and that fund our senior citizen homes.

To my mind, good governance has always been about sequencing. It means waiting until the European Union rules are there and settled upon before we build Irish machinery to enforce them. It means protecting the jobs of the State and the revenue that the State actually depends upon, and not gambling them to look busy in front of our lords and masters in Brussels in their shadow. For this reason, this Bill in its current form and in this timeline is an entire mistake. The Bill asks this House to legislate before Europe has done so. It asks us to put taxpayers' funds at risk 16 institutional months ahead of its own purpose. It firmly asks this economy to carry and absorb the cost at haste, and the firms that deliver so much of that tax revenue to us. Currently I cannot support the Bill. I ask the Minister of State to rethink it before she brings it to the House.

Deputy Richard O'Donoghue

AI in some instances scares me and in some instances it is good for different parts if it is used correctly. How many people go on their phone and then no longer know what is actually factual or not? My concern about AI is that we need to get on top of it fairly quickly. I was in the Central Bank on Tuesday for a visit with the budgetary oversight committee. There are concerns about AI in this country. Now we do not know if the Strait of Hormuz is open or not because we do not know what to believe on this or not. What kinds of decisions are being made when things are happening around this world and people have a knee-jerk reaction to something they think is happening but it is not? From that point of view, we need to do something with AI.

Children today actually look at their phones and everything they look at on their phone actually works on them whereby anything they click on is put in front of them every time they turn on their phone. Every time they flick up on something or look at something, regardless of what it is, it keeps a rota going on with that system because the algorithm knows this is what the child is interested in. So, what does it do? It brainwashes them. If you keep seeing the same thing over and over again, then you start following different paths.

When it comes to our children and our students, they go to school to learn but they now have systems put in place to see if they are using it for their homework. Now we have to put a system in place to see if somebody is doing their homework or not, and that is based over the four or five years that they are in school. They are actually using the system now to see if the are actually using AI as part of their homework system. Again, technology is good and it can help with research and so on, but now we have to put things in place to make sure it is actually the person who is doing the work. This does not just apply to children and students. A lot of people use it in their jobs at the moment and also the people in this House. How many of us are using AI? How many used AI to do their speeches today? I did not.

Deputy Joe Neville

A pen and paper.

Deputy Richard O'Donoghue

Good man, but paper does not mean that you have got it in front of you. If you have the pen in front of you, that is what makes the difference.

A lot of the time I do not come into this House with paperwork because I do not want it to be shown that it was AI-generated. What you are looking at when you look at me is me. What comes out of my mouth is what I researched myself. I have never used AI.

Something was sent to me during the week and the person who sent it said, "Richard, you looked great at a function". I was not even there. It was another me. This is the part that scares me. In a situation where we might have a terrorist issue or a financial issue, my concern is that we will have a knee-jerk reactions to something that is not actually there. How safe is our banking network? How safe are our identities when they can be used? With the stuff that is being put out at the moment, we cannot differentiate between what is real and what is not unless it is on mainstream media. That is a concern that I have with AI. We need to put whatever procedures in place to protect us, livelihoods, businesses and to stop people misusing AI for the wrong reasons.

There is no doubt that it will replace jobs. People read comments on social media. Most of the comments that are posted, if person looks at them and segregates them, are coming from bots. They are not coming from people; they are coming from bots. How many people were caught out by something that was AI-generated and put online? They then apologise and say that it was not them and was generated through AI.

When people look in at this House, they want to see real people delivering real results that they have researched themselves and that it are not AI-generated. That is my concern for the future and the here and now.

We are putting measures in place to try to counteract this, but we are also last on the field. The United States and China are on the field a long time. We are only playing catch-up. We are so far behind in stopping this from becoming destructive instead of something that can help. We are so far behind on that. We need to get on top of that for the protection and the governance of this country and all walks of life. If we are not careful, AI will take over and the simple part of our existence will be mirrored everywhere we go. I would like to see how AI will get on with me because sometimes I do not know what I am going to do myself when I walk out the door, so it will have a hard job with me. There are, however, a lot of people who are repetitive and it will screen them and know their next move all the time.

We need to put measures in place for the protection of everyone and the protection their finances from criminality so the sooner we get something put together the better. We also need to get it right. It needs to be researched correctly - not AI research. It needs to be researched properly for the protection of this country.

Deputy Joe Neville

Deputy O'Donoghue just spoke, and I have a west Limerick background. When I heard about AI when I younger, it was on the farm, and it was not artificial intelligence. It was artificial insemination, but things have changed over many years. That joke has been made many times but it always speaks to me about how we have to keep in touch.

Deputy O'Flynn spoke earlier, and I have a lot of the same concerns he has. He spoke clearly about them, and many people have the same concerns. We need regulation but we have to be careful. We do not want to over-regulate or over-instruct, but AI is an evolutionary change. We cannot have it becoming like the Wild West, so we need strong rules and regulations in place, and we need to harmonise them and make sure they are there because they have to be. That is why this is such a balancing act and why it is something that is brand new.

We all know that artificial intelligence is having an effect on all our daily lives in many different ways. Some people use it all the time. Some people use it for work. It is changing small companies, and it is really having an impact on large companies. Employee numbers are probably smaller. I started in PriceWaterhouseCoopers as a young auditor. I am sure that within the next couple of years, those numbers will be reduced over time, especially in the US. It will be first to embrace this and after that Ireland will, no doubt, follow the same path. Companies will no longer need people to go in and assess the basics. They will go in at a higher level.

Regarding the impact that AI will have, we do not know where it is going to end up. Those large companies that are in charge of it are working with other companies and getting in deeper. Ultimately, the risk is that people will be excluded and that is the larger existential AI risk in the workplace.

Previous speakers mentioned about a potential AI tax, a term I hate, on users and corporates, which is something I referenced before, because no one knows where it will end up. If workers, companies and smaller SMEs are going to be displaced then we have to have some way to monitor this. That is further down the field, but these conversations need to be had at the outset because we do not know the impact that AI will have. That is why I welcome the Bill in place and the office being established. We need someone to spot the gaps, differences and potential impacts because this needs to be monitored.

As was referenced earlier, Ireland is currently benefiting from AI through corporation taxes and even business. I was in with the Central Bank yesterday. Its officials spoke about how a vast element of the growth in the economy is related to AI and data centres. That is where we are seeing the benefits in many ways that other countries are not. I am not going to speak out of two sides of my mouth and complain and worry because we are seeing a real benefit from it. At the same time, it does not mean that we cannot look towards the future.

I welcome the Bill, and it is really important that we do this. We also need to ensure that we do not over-regulate and go too far. At the same time, we have to monitor it and that is the balance that we will have to get right out of this.

Deputy Keira Keogh

If Deputy O'Donoghue thinks that he has not used AI before, he definitely has. Even if he has not used a particular AI app, if he put a search into Google this evening, he would have got an AI summary whether he wanted it or not.

As a member of the Oireachtas Joint Committee on Artificial Intelligence, I welcome this Bill and the work of the Department officials and the Minister of State. It is legislation that is hugely important and necessary to ensure that Ireland meets its obligations in overseeing the development and deployment of artificial intelligence systems. More importantly, this legislation aims to protect our citizens and uphold their fundamental rights. I welcome the establishment of the AI office, but I would like it to have more of a role in advocacy and literacy. It has the opportunity to be a one-stop shop with a cross-sectoral approach that will look at health. We know that people are turning to AI at the moment to get advice on health. Children in our education settings are using AI and they may not be aware of the hallucinations and the misinformation that it can produce. We have the risk to our SMEs and other jobs in business and enterprise. The disability space has a real concern that the bias that is already in place will unintentionally be reinforced by AI. The AI office could be a trusted one-stop shop to also engage in advocacy and literacy. I welcome the enforcement mechanisms and sanctions. It seems the Bill will give us robust oversight and significant penalties.

As the chair of the children and equality committee, I must examine this through the lens of children's rights, and I see where there might be gaps. Under the heading "fundamental rights", children will be protected but the issues relating to technology and social media have more consequences for our young people growing up in this new reality.

I have serious concerns about harmful chatbots. We know there have been times when people have been coached into suicide. I often talk about the fact that there was no Tinder when I was in secondary school and now there are Tinder babies and marriages. I am worried about the next generation. It might not even be Tinder. They might have their first relationship with a chatbot. What happens then when the company introduces a significant subscription fee for that chatbot?

Do they then lose access to their boyfriend or girlfriend, and where does that leave their mental health? I think children will face this concern more than others. We know they are facing deepfakes and having their videos manipulated for potential bullying. I ask that provision be made for the expertise of children and young people, perhaps through an advisory panel, which we have seen work so well in other areas, and to support the rights groups that are watch-guards in this space. This afternoon I was on a panel discussion with the Children's Rights Alliance, and this is something it and its members are passionate about. I believe there will be scope to address those concerns on Committee Stage or at phase 2 in the autumn. Children need named protection in this Bill. As I said, AI features are now embedded into almost everything they do, from daily Internet searches and their gaming to the educational tools within their classrooms. On every single screen a child touches, AI is now a feature.

The fundamental rights agencies will, of course, still play a role. They can make complaints, they will co-operate with regulators and they may channel decisions through legal mechanisms. We must continue our work towards a system that genuinely protects the rights of all those who might be most at risk. I have mentioned children in my role, but the disability community and our older people are also very much at risk of AI, aside from the jobs that could be lost in the enterprise sectors, an issue which has been very much covered.

As we approach the beginning of Ireland's Presidency of the Council of the EU, I really welcome this Bill and our obligations under the AI Act. I hope we remain steadfast as we continue to demonstrate our commitment to online safety in promoting responsible AI innovation.

Deputy Barry Ward

I welcome the Bill. Regulation in this space is perhaps more important than any other regulatory Bill we will pass in this House this year. Any new technology poses problems, difficulties, challenges and threats and, undoubtedly, that is true of artificial intelligence. It does not mean we cannot regulate to protect ourselves and it does not mean we cannot put measures in place to protect ourselves and our children. There is a concern about the notion of over-regulation, that we will therefore stifle innovation or prevent people from coming up with new ideas or new technologies. In fairness, artificial intelligence represents one of the greatest achievements of mankind from the point of view of creating something that can grow itself, feed itself and inform itself. That in and of itself should not be overlooked. The danger that comes with that is that we do not control how it feeds itself or how it informs itself. We have already seen the many pitfalls where artificial intelligence will throw up results in a search online or through an artificial intelligence app that are patently false. As a lawyer, I can tell the House that I have seen many instances of that happen in the legal world, thankfully not in this jurisdiction but certainly, if the Internet is to be believed, and frequently it is not, quite a lot in America. You see lawyers in legal cases submit legal submissions which cite cases that never actually occurred. The danger is that artificial intelligence does not have the ethical framework that humans have. It does not have the understanding. It certainly does not have emotional intelligence or any of those things that we factor into human decision-making.

We therefore need to put in place regulation to ensure we do not have a raft of falsehoods being propagated by artificial intelligence. Now, that already happens. It happened with the Internet before artificial intelligence. It happens very often at the nefarious hand of humans with normal intelligence, but artificial intelligence can do it on a scale and at a pace we cannot match, so regulation is hugely important there.

I do not believe that regulation stifles innovation necessarily. It can, of course, and over-regulation can hold things back. However, one of the things we can say very proudly in Europe is that innovation and regulation are not mutually exclusive; we can have both. We can have regulations that protect us, protect our information systems and protect our children from exposure to these things online, but we can also have an environment that allows entrepreneurs, developers and the creators and the creative people in this world to constantly build new technologies. That is a good thing. Many of the criticisms that are made against artificial intelligence were made against the Internet and the computer and, before that, the machine that sliced bread, if proverbs are to be believed. I am not a Luddite. I do not believe we should not be progressing - of course we should - but we need to progress in a way that can ensure the protection of the citizenry's rights and the integrity of what they view, for example, on the Internet. Reference was made earlier to the fact that anybody who has done a Google search in recent times has used AI, whether they actively consented to it or not. I have an issue with that because nobody asked my consent to enter my search criteria into an AI bot or an AI application. This is the kind of thing for which regulation is needed. If people are to engage with artificial intelligence, they should do so in an informed way and actively consent to doing that. Otherwise they do not really know what is coming back or whether they can rely on the search results that come back. Perhaps what is most important is that, irrespective of what regulations arise from this legislation, people still have to check things themselves.

May I make just one point about the drafting? The Bill does not actually define what artificial intelligence is. In section 2, the interpretation section, it does make reference to the European legislative instruments, but it does not actually say what they say. That is a drafting difficulty that means people will have to look up that stuff elsewhere, which can be quite difficult for them.

Minister of State at the Department of Enterprise, Tourism and Employment (Deputy Niamh Smyth)

I thank everybody for their contributions, their helpfulness and their thoughtfulness in their statements, comments and observations. It has been a very wide and varied discussion. AI inevitably touches almost every aspect of society, from innovation and economic growth to public services and jobs. While these are important and wide-ranging issues, today we are here to introduce this legislation, which is focused on implementing the EU AI Act in Ireland and establishing the framework necessary for its effective supervision and enforcement. I have no doubt this House will continue to discuss and to debate the broader implications of AI in the months and years ahead. That is both necessary and welcome.

There were some comments that this is very rushed. Since I have come into the Department 16 months ago, these very smart and bright people to my right, the officials in my Department, have been working night and day on this legislation. They could in no way be accused of having rushed this without real and proper consideration and thought. While I do appreciate we are bringing it through the Houses quite quickly, I am certainly open to working with the Opposition on any aspect of this in the weeks ahead.

On the protection of fundamental rights, which came up quite a bit across the House, and the enforcement regime in the Bill, I point to the fact again that the EU AI Act has direct effect in Irish law and it sets out very clear obligations on providers of AI systems, deployers, importers, etc., that the Bill must be read alongside the EU AI Act. You cannot read or consider one without the other. Safety, ethics and fundamental rights are the key focus of the AI Act. The enforcement regime in the Bill is robust and builds on the strong market surveillance frameworks already in place. It enables the ramping up of enforcement, from contravention notices to encouragement of compliance to the larger fines. That is incredibly important to say about safety, ethics and fundamental rights.

There was some discussion about the labour market. I would be the first, and I think I always have been since I have come into this Chamber to discuss artificial intelligence, to put my hand up and say there will be labour disruption and displacement. There is certainly no putting our heads in the sand over that matter. The establishment of the AI office seeks to address that. The Government is committed to preparing Ireland for the inevitable transition that is happening within the workplace. Ireland has had its third highest level of STEM graduates. One of the biggest demands of qualified students coming here to Ireland is for AI-related jobs, even though I acknowledge there have been job lay-offs in certain companies. However, it is important to note that there is still huge and driving demand for those skilled in AI. The leading providers have established in this country, along with our strong education system, and we continue to attract that talent and investment. It was noted that we have some of the biggest AI companies here in Ireland. That is because we create the environment for that. We are in no way attempting to compromise that or compromise any of the jobs that are directly affected by that. The national AI and digital strategy provides establishment of OBAIR to monitor and provide insight into the evolving impact of AI on skills and the labour force. That is all done under the Department of Further and Higher Education, Research, Innovation and Science. Within our own Department, we are also setting about to establish an observatory that looks at exactly that - the implication for jobs - and trying to mitigate that as much as possible.

The national AI digital strategy has been refreshed on a number of occasions to try to really be that agile blueprint that we can refer to and work with all the time. It is certainly not going to be stiff or stale, or sitting on a shelf. Again the officials are to be applauded for the incredible work on that.

Deputy Lawlor went into great detail on constitutional matters. I know he may not have been here for my introduction to this but all the provisions of the Bill have been drafted with the Office of the Parliamentary Counsel, with constant engagement and consultation with the Office of the Attorney General on the legal and constitutional matters. However, I have taken copious notes, and I am sure my officials have too, on the matters he raised tonight. We will certainly do our best to bring peace of mind to that.

Another matter that was raised on a number of occasions was AI literacy. The role of the AI office is not just oversight of the regulatory market surveillance and regulatory bodies that already exist. As the general scheme recognises, effective AI regulation depends not only on enforcement but also on the appropriate level of AI literacy of those who develop, deploy, use or are impacted by AI systems. To support the development of AI literacy, the AI office has been given a very specific role around AI literacy, supporting awareness of AI-related risks and obligations; providing guidance and information to organisations and to public bodies; facilitating informed engagement with AI technologies and contributing to public trust in terms of the AI systems. It is not going to be a one-trick pony. It is a very broad, all-encompassing, across-the-board piece of legislation, which has been described to me as horizontal, that the AI office will bring to the fore.

I thank Deputy Conway-Walsh for being here until the conclusion of the Bill. She raised certain things and I assure her they are being addressed. I can provide her with further reassurance and notes. She raised independence and resourcing. I am certainly going to address that. I think I have addressed displacement already in my closing comments.

On regional developments, the Deputy is absolutely right that there is huge potential. She and I both come from very rural parts of this island, and I think that is one of the absolute benefits to AI deployment.

Simplification has come up quite a lot, with negativity about having another regulatory body. The Department and officials have been so cognisant of not bringing in more red tape or burying business and SMEs in more bureaucracy, to actually make things simple and not add another layer of time-consuming cost burden to business. I want to acknowledge that and state it on the record.

Deputy Conway-Walsh also raised transparency and the creative industries. As somebody who has come from the creative industries, I am very mindful around copyright and artists' work being mimicked or copied without their approval, their say or any remuneration. All of that is being taken into consideration and I can supply the Deputy with further information on all the matters she raised here tonight.

I hope I have addressed Deputy Keogh's piece on AI literacy in some way tonight, too.

I thank all the Deputies for their contributions to the debate and look forward to their support in this Bill and its further consideration on Committee Stage. I re-emphasise that the purpose of the Regulation of Artificial Intelligence Bill 2026 is to provide for the implementation in Ireland of the AI Act and EU regulation. This Bill is a landmark for Ireland's digital regulatory framework. AI is a transformative technology which offers extraordinary potential for our economy and our citizens, but realising that potential requires appropriate oversight and accountability. This Bill delivers exactly that. It fulfils Ireland's EU obligations, giving effect to Ireland's pioneering AI Act in domestic law by ensuring we have the national infrastructure in place to supervise and enforce it effectively.

The Bill establishes Oifig IS na hÉireann, the new AI office in Ireland, as an independent institution at the centre of our AI regulatory system, and empowers our competent authorities with the investigative and enforcement tools they need and, most importantly, the skill base that everybody has referred to, the skills that are needed in terms of scrutinising AI that will be available to the competent authorities.

Ireland has adopted a distributive model for the regulation of AI, levering the expertise and capability of the State's existing regulators and providing that expertise to them. This is speaking to the fact that we do not want to bring any extra regulatory burden. We must remember that regulation is the gold stamp of approval. The businesses I have talked to want that approval. They want to know what the safeguards and rules of engagement are around AI. That is going to bring confidence to this country, particularly to those in business and enterprise. It is not mitigation or putting a huge emphasis on enterprise and forgetting about the human centrepiece, which is being really considered throughout this work.

I would particularly like to thank the many Government Departments and their authorities, agencies and offices that have been involved in consultation on this Bill. My Department remains actively engaged with the competent authorities and their line Departments as preparations for implementation of the AI Act and enactment of the Bill continue. Again I note that the Bill is a technical instrument that will ensure that the rights and protections provided by the EU's AI Act will be rigorously asserted in Ireland for the benefit and protection of Irish citizens. It does not add to or alter the obligations on businesses under the AI Act. It is mandatory under the AI Act for EU member states to give effect to these national provisions by 2 August 2026. I thank all the colleagues around the House and most particularly the officials in the Department, who have worked might and main to bring this to fruition.

Question put and agreed to.

Dáil Éireann debate - Wednesday, 24 Jun 2026

Regulation of Artificial Intelligence Bill 2026: Second Stage

Dáil Éireann debate -
Wednesday, 24 Jun 2026