Joint Committee on Enterprise, Tourism and Employment díospóireacht

General Schemes of Regulation of Artificial Intelligence Bill 2026 and Data Bill 2025: Discussion (Resumed)

An Cathaoirleach

Before we proceed, I have a few housekeeping matters to go through.

I wish to explain some limitations to parliamentary privilege and the practice of the Houses as regards references that witnesses make to other persons in their evidence. Witnesses are protected by absolute privilege in respect of the presentation they make to the committee. This means that they have absolute defence against any defamation actions for anything they say at the meeting. However, they are expected not to abuse this privilege. It is my duty as Chair that this privilege is not abused. Therefore, if witnesses' statements are potentially defamatory in relation to an identifiable person or entity they will be directed to discontinue their remarks. It is imperative that they comply with any such direction.

I advise members of the constitutional requirement that they must be physically present within the confines of the Leinster House complex in order to participate in public meetings. I will not permit a member to participate if they are not adhering to this constitutional requirement. Therefore, a member who attempts to participate from outside the parliamentary precincts will be asked to leave this meeting. In this regard, I ask any member partaking via Microsoft Teams that prior to making their contribution, they confirm they are on the grounds of the Leinster House campus.

Members and witnesses are reminded of the long-standing parliamentary practice to the effect that they should not criticise or make charges against any person or entity by name or in such a way to make him, her or it identifiable, or otherwise engage in speech that might be regarded as damaging to the good name of the person or entity. Therefore, if their statements are potentially defamatory in relation to an identifiable person or entity they will be directed to discontinue their remarks, and it is imperative that they comply with any such direction.

I propose that we publish the opening statements and submissions provided by the witnesses on the committee website. Is that agreed? Agreed.

With regard to speaking arrangements, witnesses will be invited to speak for approximately ten minutes. I will then allow members to ask questions and make comments for approximately seven minutes. We will only have second round of questioning if time permits. If we do have a second round, members will be allowed to ask questions or make comments for approximately four minutes. Members may be called as they appear on the week 3 speaking rota, which was circulated before the meeting. Committee members may substitute within their parties or groups. Non-members or substitutes may only speak after committee members or substitutes have spoken. Are these arrangements satisfactory and agreed? Agreed.

The minutes of the meetings that took place on Tuesday, 28 April, Wednesday, 29 April and Wednesday, 6 May, have been circulated. Are they agreed? Agreed.

We turn to the pre-legislative scrutiny of the general scheme of the data Bill 2025 and the general scheme of the regulation of artificial intelligence Bill 2026. The committee decided to carry out pre-legislative scrutiny of these general schemes. We have had a large response to our call for submissions on these general schemes. I acknowledge and thank those who made submissions to the committee.

I welcome the witnesses. From the Irish Human Rights and Equality Commission, we are joined by Mr. Liam Herrick, chief commissioner, and Ms Rebecca Keatinge, head of monitoring and compliance. From the Children’s Rights Alliance, we are joined by Ms Noeline Blackwell, online safety co-ordinator, and Ms Alex Murphy, online safety legal officer. From Digital Business Ireland, we are joined by Mr. Victor Timon, chair. I invite Mr. Herrick to make his opening statement.

Mr. Liam Herrick

I thank the Cathaoirleach and members of the committee for the invitation to attend. We welcome the committee's focus on the general scheme of the regulation of AI Bill. The Irish Human Rights and Equality Commission is Ireland's independent national human rights institution and Ireland's independent national equality body. We are also designated as a fundamental rights body under Article 77 of the AI Act. I will confine my comments to the regulation of AI Bill.

In October last year, I appeared before the committee on artificial intelligence to set out our concerns on the risks associated with an increasingly loud anti-regulation discourse on the question of artificial intelligence, a discourse presented in the name of competitiveness but which comes with a significant cost in terms of fundamental rights protections. As I said to that committee, we are at a transformative moment where governments and policymakers seek to regulate a force that is already reshaping economies, governance and fundamental human rights. Ensuring effective protection of rights, including children's rights, workers' rights and the rights of people with disabilities is a great challenge for us all, particularly given the influence of the large technology companies on shaping narratives regarding AI and pushing for reduced accountability and reduced liability for AI-related harms.

Over recent months, we have seen several stark examples of how AI can impact on human rights. The used of AI-generated deepfakes to spread misinformation at a key moment in the run up to the presidential election and the use of AI to generate sexually explicit images of women and children are just two notable and disturbing examples of that type of technology in recent months.

At the policy level, we in the Irish Human Rights and Equality Commission have been concerned at proposals to amend the recording devices Act to allow for retrospective use of biometric technologies, including facial recognition technology. The risks associated with AI do not only come from bad actors, but even well-intentioned initiatives can carry unintended consequences. It is difficult to find precedent for the potential of this technology to fundamentally transform society, how we interact with the State, with our services and with each other. However, for the public to have confidence in AI, human rights standards and effective regulation are critical. This means we need to see clear rules, enforceable standards, effective regulatory structures and accessible mechanisms for accountability. Without such public trust, the transformative potential of AI will not be fully realised.

We believe the committee’s examination of the general scheme before it today marks a significant moment in the implementation of the EU AI Act. It marks Ireland’s first steps in regulating AI at national level and presents a real opportunity to ensure that forthcoming legislation includes strong and effective fundamental rights and equality safeguards. I first address proposals with regard to the AI office. Central to rights and equality safeguards and the development of a coherent and responsive regulatory framework will be the operation of an independent and effective AI office in Ireland. While we welcome the proposed establishment of this office under the legislation, we are concerned that the provisions proposed in the general scheme will constrain the office’s operational independence. These constraints include provision for broad ministerial powers to issue directions to the office, a lack of transparency on the appointment and removal of the CEO of the office and restrictions on the CEO's ability to comment on Government policy. We recommend strengthening the role of the office in protecting fundamental rights and ensuring the adequate participation of groups disproportionately impacted by AI.

Another aspect of this general scheme which the commission reviewed in its submission was on the role of Article 77, or fundamental rights bodies, which is an absolutely crucial part of the regulation of AI. We are of the view that the provisions set out in the general scheme on how the article 77 bodies, of which we are one, are to operate are disappointingly limited and lacking in ambition. Without a stronger mandate and support, we believe the nine designated fundamental rights bodies will not be able to play a meaningful role in the regulatory framework. For instance, the general scheme does not make provision for the resourcing that is required to enable the Article 77 bodies to function fully and effectively as regulators. There are also inadequate structural co-operation and consultation mechanisms provided for in the general scheme, and there is no provision for the bodies to have access to relevant technical expertise. We are of the view that these gaps will need to be addressed through express legislative provisions. We note that such measures have been provided for in the general scheme with regard to market surveillance authorities, and we believe that equivalent provisions should extend to Article 77 bodies.

With regard to the EU database, central to the work of Article 77 bodies in building public confidence in adequate safeguarding is the availability of publicly accessible, transparent and complete data on the uses of AI, particularly within public services. We note the establishment of the EU database but we also note that it has substantive limitations and the exclusions relating to it. Among the latter is the use of AI in law enforcement, migration and border control management.

We also believe that several forms of AI which could be captured include those relating to public bodies providing public services. Due to these and other limitations, we recommend provision in domestic legislation for the establishment and maintenance of a national register of AI systems in use by public bodies and in the delivery of public services. We believe this would be consistent with the commitment to transparency and accountability underpinning the EU AI Act and Government policy on AI.

With regard to the digital omnibus provisions, it is stated in the general scheme that it is intended that there will be further changes to respond to the EU simplification proposals. We reiterate our concerns that the omnibus on AI proposed by the Commission will, if it is formally adopted, delay oversight, monitoring and transparency. We call on Ireland to move beyond minimum standards at EU level by preserving the key existing fundamental rights protections and timelines.

The conclusions drawn in our observations on the legislation point to the need for strong safeguards. To be effective, AI governance must protect fundamental rights and ensure legal certainty. Under-regulation will harm individuals and democratic systems. The challenge for our legislators in the Oireachtas lies in designing adaptive, forward-looking frameworks that are capable of evolving alongside the technology that we seek to govern. The landscape reflects a maturing public understanding of AI’s societal impact. I commend the crucial role being played by all Oireachtas committees, including this one, in advancing democratic consideration of these issues. The direction of travel, we believe and hope, is toward principled, risk-based and human-centric regulation, supported by strong institutional frameworks and international co-operation. The success of this approach will ultimately depend on our ability as a country to remain agile and ensure that we are responsive to technological change, grounded in democratic values and effective in practice.

An Cathaoirleach

I thank Mr. Herrick very much. I appreciate that contribution. I invite Ms Noeline Blackwell to make her opening statement.

Ms Noeline Blackwell

I thank the committee for the inviting us to contribute to these deliberations. We sent the committee our submission on the general scheme of the artificial intelligence Bill a few weeks ago. We really welcome the opportunity to speak to the recommendations we included in it for the committee's consideration.

While we have spoken elsewhere about the impact, positive or negative, that recent generative iterations of AI have had and that future iterations will have on children and young people, the committee will see that our submission concentrated on certain matters which are entirely within the power and competence of the Oireachtas. We note that the proposed legislation under discussion will be the vehicle for the implementation of the EU-wide scheme of regulation of AI. From that perspective, it is extremely important. No matter how AI changes, this legislation will be our national vehicle for implementing EU law on the matter in Ireland, and it is important that it contains mechanisms that are not only efficient but that also respect, protect and promote the rights of the people of Ireland and, from the Children’s Rights Alliance perspective, the rights of the children and young people of Ireland.

We chose to concentrate our submission on three vital parts of the legislative machinery where we feel the general scheme appears to fall short of the protection that children and young people need and deserve. They are: the recognition of and provision for the implementation of the rights of children and young people, where there is a gap; the need for effective independence and impartiality of the AI office; and the adequate provision for fundamental rights oversight and protection. The committee will see that there is some overlap between the issues we are speaking about and those that the Irish Human Rights and Equality Commission spoke about.

Where there is a specific gap is in specific recognition and provision for the implementation of children's rights. In a criticism that we echo in relation to the EU AI Act more generally, the general scheme of the Bill fails to recognise children as a vulnerable category and, more importantly, fails to ensure that measures adopted explicitly protect children’s rights. We know that such protection is in line with current Government policy, the policy framework for young people, Young Ireland, and the newly launched digital and AI strategy for the period 2025 to 2030. However, that is not clear from the general scheme of the Bill. The law on and practice regarding AI are likely to continue to change at speed. We propose that the committee consider a structural safeguard for children’s rights, which is the inclusion in the Bill of an obligation that the legislation commits to requiring any and all measures of law and policy to be subjected to general frameworks that can apply in almost any situation. As is recognised in the Young Ireland policy framework, a child rights impact assessment is an established, known and useful tool. We submit that this tool could first be applied to the eventual AI Bill, as well as ensuring that measures, functions and policies that arise from the legislation be similarly viewed through a child rights impact assessment lens. Similarly, the general scheme is silent about the participation of children and young people in decision development and decision-making. Children and their voice should be considered at all levels of decision development and decision-making, including by any regulatory sandboxes and the board of the AI office.

Turning to the AI office, the EU AI Act requires a high standard of independence and impartiality of national competent authorities to safeguard the AI office's unbiased and objective activity and enforcement. The general scheme notes that the AI office should "operate independently in performing its functions, ensuring impartiality and integrity in its role". This is a long-term obligation of the AI office. The general scheme, however, contains measures, as Mr. Herrick already said, which could, over time, fundamentally impair the essential independence and integrity of the office. For instance, the general scheme proposes powers for the relevant Minister to give written obligatory directions to the office. In addition, it proposes that the Minister shall appoint the chairperson and members of the board according to subjective criteria, which carries an inherent risk to the independence and impartiality of the office.

In another aspect of the general scheme, still on the AI office, it is to be given powers facilitating and enabling co-operation between competent authorities in Ireland and internationally in promoting AI innovation and literacy, and in facilitating access to technical expertise for national competent authorities. However, there is no provision that we can see to require the office to facilitate co-operation between the nine designated agencies for the protection of fundamental human rights. The functions of the AI office will be well placed to facilitate this and enable those agencies, and we submit that the Bill should be amended to require it to do so or that alternative provisions be put in.

Moving to the issue of fundamental human rights authorities, while Ireland does seek to be seen as a leader in innovation, it has committed, in its digital and AI policy, to the upholding of fundamental human rights in the context of AI. However, the general scheme of the Bill fails to give this adequate attention. Some of the nine authorities already designated by the Minister as fundamental rights authorities have a focus primarily on human rights and equality. The Irish Human Rights and Equality Commission would be one of those. Some are regulators, including Coimisiún na Meán and the Data Protection Commission. Some are also designated competent authorities. While there is a wide range of skills and competencies in the nine organisations, in the current general scheme, it is just not possible to see how and if they are all to work together, how and if they are all expected to deal with the full range of fundamental human rights or, equally, how competent authorities are to engage to protect fundamental rights and the other way around. Furthermore, there seems to be no particular funding or co-ordination structure to allow those authorities to fulfil their purpose to act as guardians of fundamental rights and a safety net for the people of Ireland in general and for children who are recognised in the EU AI Act as particularly vulnerable in the context of protecting them from the harmful impact of AI.

We recommend that the Bill and associated legislative policy provide for clarity, co-ordination and adequate resources to allow these authorities to carry out their mandates.

My colleague, Ms Murphy, and I are happy to engage with the committee in whatever way may be useful to you, a Chathaoirligh.

An Cathaoirleach

Thank you very much, I appreciate that. Our next contributor is Mr. Timon. I invite him to make his opening statement.

Mr. Victor Timon

I thank the Chairman and members of the committee. I am a partner with Byrne Wallace Shields LLP and am here in my capacity as chair of Digital Business Ireland. I thank the committee for offering DBI this opportunity to present our views on the general schemes, especially the AI scheme.

Digital Business Irelandis the leading and most active voice for Ireland’s digital and e-commerce sectors. It represents a community of over 8,000 businesses from multinationals to Irish SMEs, as well as experts in law and academia. Digital Business Ireland’s mission is to accelerate the digital transformation of businesses in Ireland and drive the growth of Ireland’s digital economy. DBI is a member of the Government’s enterprise digital advisory forum, which I sit on, and the retail forum, and represents Ireland at European level as a member of Ecommerce Europe.

Last month, DBI published a report, Making AI Work for Ireland, which set out the enormous opportunity AI offers to enhance productivity, unlock new capabilities and generate real value across Irish businesses and public services. Moreover, the report makes the case that a better balance must be struck between digital regulation and supporting businesses to innovate and grow. With this in mind, Digital Business Ireland welcomes the publication of the general scheme of the regulation of artificial intelligence Bill, but we believe the Bill could be further strengthened and enhanced to actively support the effective, responsible and innovative adoption of AI by businesses in Ireland.

The general scheme comprehensively addresses issues of compliance and enforcement, which are about protecting people and their rights, but we are concerned that of the 111 heads forming the general scheme, 73 relate to enforcement-related matters, with only head 35, which deals with AI regulatory sandboxes, being related to business support. DBI believes the general scheme should be rebalanced to add more economic objectives and provisions relating to support for businesses and innovation.

In March 2024, the DBI European election manifesto first called for the establishment of a national AI office to provide a focal point for the implementation of the EU AI Act in Ireland. We subsequently welcomed the Government’s plan, announced as part of the new Action Plan on Competitiveness and Productivity, to establish a new AI office in Ireland. In welcoming the proposed new office, and since then, DBI has been consistent in our view the AI office must have a clear mandate to support businesses in Ireland and the growth of Ireland’s digital economy. In our written submission to the committee, DBI recommended amendments to the proposed Bill to strengthen the mandate of the AI office in Ireland. We recommended adding to the functions of the office set out in head 9, backed up by further detailed sections, to specifically reference the promotion of AI innovation and adoption, the fostering of AI literacy, the provision of guidance on regulatory requirements and compliance support, and the provision of training to businesses and the workforce.

Looking beyond the AI office, DBI believes consideration of this Bill provides a timely opportunity to review and update the mandates of other Irish digital regulators, especially the Data Protection Commission and Coimisiún na Meán, to make them actively support compliance with new AI rules and to promote innovation and economic growth. This would not displace or diminish their independent regulatory functions but rather complement their enforcement role by actively supporting businesses in achieving compliance with digital regulation, including in relation to AI.

In regard to the distributed regulatory model, while the general scheme establishes a framework of co-operation between multiple competent authorities, it does not clearly regulate how conflicts or gaps in enforcement should be resolved. DBI would like to see clarity on the process of determining which authority should lead where more than one seeks to investigate the same entity. Head 18 of the general scheme sets out the proposed membership of the board of the AI office. DBI believes industry associations, not just our own, would be in a strong position to nominate highly qualified individuals who would bring a depth of business experience and insights on the interplay of regulation, compliance and business support to the role of board members. We suggest the Bill requires at least two representatives from recognised industry associations on the board to ensure the AI office maintains a strong economic and innovation focus.

The general schemes of the regulation of artificial intelligence Bill and the data Bill provide for the funding of competent authorities by way of levies. DBI is concerned that in the context of the substantial cost-of-business challenges facing businesses of all sizes in Ireland, the potential imposition of new sets of levies could disproportionately impact small indigenous businesses as well as undermine the competitiveness and attractiveness of Ireland as a leading global centre for digital business and innovation. Digital Business Ireland would like to see stronger provisions in the Bill to ensure a fair and proportionate model for the calculation of levies and we would generally urge the costs of any levies are kept to a minimum. We add that if businesses are expected to fund regulators then it is only fair and reasonable the same regulators should have a clear mandate to support businesses in Ireland and the growth of Ireland’s digital economy.

I thank the committee again for allowing me to represent Digital Business Ireland. I look forward to discussing the general schemes with members.

An Cathaoirleach

That concludes the opening statements. We will now go to members for their contributions.

Deputy Tony McCormack

I thank Mr. Herrick, Ms Blackwell and Mr. Timon for their opening statements and the submissions they made to us previously. I compliment Ms Blackwell on her great work in human rights and children's rights over the years, which is very much appreciated. We can see that if this area is not legislated for properly it could have huge and devastating effects on society in general. Looking at the way social media has developed over the years from being something that was very much a good thing to where it is today, the bad outweighs the good a lot. When we take that into consideration, we have to be very careful what we do with AI legislation.

As Fianna Fáil's spokesperson on small- and medium-sized enterprises and retail, my questions are going to be on the wider retail sector and ensuring we strike the right balance between regulation, innovation and competitiveness. With that in mind, my questions are directed to Mr. Timon and Digital Business Ireland. Its submission raised concerns about proposed levies on data-processing services. Will Mr. Timon outline the likely financial impact on SMEs and indigenous businesses of this legislation?

Mr. Victor Timon

We do not know what those levies are going to be so we are trying to encourage that they not be excessive. They should be proportionate, especially for SMEs. Larger companies may be able to afford higher levies, but this should certainly reflect the pressures SMEs are under currently due to costs and keep those to a minimum. With those levies, we would like to say we were paying for something that is positive. It is not just about enforcement but about getting the regulator to use those funds to help with education, skills training and things like that.

Deputy Tony McCormack

AI can be very positive but there are obviously the dark sides and the nefarious players that can get involved. We have seen what has happened with the big tech companies over the years and, as I said, social media and what has happened there with deepfakes, misinformation and what have you. The law we bring in has to be robust enough that it can deal with those issues but also has to be flexible enough that Ireland's businesses are not constricted and lose competitiveness as a result of it. As we move along things will change and we have to be flexible enough to change those laws.

Mr. Timon referred to the need for a "fair and proportionate" levy model. What would that look like in practice, especially for smaller firms that may already be facing significant compliance costs?

Mr. Victor Timon

I cannot really put a cost on it because I do not know what Government has in mind, but we would expect it to be something that was bearable and not another major cost that discouraged people from being involved in AI. When we look at the Irish economy, we see how little money, based on surveys we have done, people are prepared to spend on implementing AI because they have so many other costs.

We found that 60% of firms were going to spend less than €10,000 this year on AI. We are trying to encourage Government to make them spend more by increasing grants to them and so on. To impose high levies on them would be a mistake. The Government also has a committee on how to reduce costs for businesses. It would probably be counter-intuitive to have that committee looking at how to reduce costs and then impose greater costs.

Deputy Tony McCormack

Does Mr. Timon feel that those 60% of companies surveyed that are spending €10,000 or less on AI is because they are taking a wait-and-see position on the whole situation? It is so new, they want to see what is happening in the market and technology changes so fast that if they get stuck into something now, by the time it is implemented in their business it is out of date and they have to go further. Is it a fear of being involved in AI or is that all they can afford to invest?

Mr. Victor Timon

It is probably a combination of all those things. Some companies do not have the funds to invest in it but they need to be encouraged by way of allowances. Others have a fear. We all know how quickly things change. It is a combination of all those things. The more we can encourage them to invest, in terms of competitiveness, the further up the ladder they are going to get. Everybody seems to be dabbling at the bottom end with very small licences. We want to encourage them to move up the ladder and really use it competitively and productively.

Deputy Tony McCormack

With his Digital Business Ireland hat on and looking after small- and medium-sized enterprises, does Mr. Timon feel there is a risk that smaller Irish firms will struggle to compete with larger multinational companies under the proposed framework? If so, what safeguards and supports should the Government put in place? I am more interested in the safeguards because Mr. Timon said he would like to see the Government come in with grants, but what sort of safeguards should we put in so the smaller Irish businesses are not at a competitive disadvantage?

Mr. Victor Timon

The thing about this technology is that it actually allows very small businesses to compete with the larger ones. We are seeing the way the whole market is being turned upside down in terms of technology with a lot of the software as a service, SaaS, companies running for cover. There is an opportunity for small businesses to get on an equal level with much bigger, global ones. It is not a question of putting in protections for those because that seems to be doing something against somebody else. We want to see support for those businesses to bring them up, rather than trying to penalise somebody else.

Deputy Tony McCormack

After all the work that Digital Business Ireland has done for the submission on this Bill, what is the biggest worry for Irish business out there with regard to the Bill?

Mr. Victor Timon

With respect to my colleagues, I share a lot of their views in terms of what enforcement and regulation should do. The Bill is missing an opportunity to include provisions on innovation, compliance guidance and that aspect for business, while not taking away from what my friends here also require from it.

Deputy Tony McCormack

I thank Mr. Timon, Ms Blackwell and Mr. Herrick.

Deputy Paul Donnelly

I thank the witnesses for their presentations. This is one of those issues that is on everyone's mind, whether you work in an industry where people are fearful of losing their jobs because of AI or you work with young people or children where there is a fear around the images being created and manipulated. The scope of AI is something people find really difficult to get their heads around. It is so new. It also not new because AI has been around a long time, but the development of AI in the last couple of years has been phenomenal.

It brings its worries and concerns because this is being driven primarily by a small number of groups and a small number of companies. Interestingly, all of them are headquartered here in Ireland. We have seen some of the worries and concerns, which I share, in relation to the protection of children and women and around international human rights, the rights of people, their rights to dignity and their right to own themselves. Somebody puts something on the Internet and all of a sudden they lose all their rights because they signed up to a particular Facebook page, or Twitter or X. The reams and reams of conditions they sign up to basically means these companies own their rights. They own all of their image rights, which is very worrying.

As we are here in Ireland and most of the major companies involved in AI are headquartered in Ireland, I have a concern, which is borne out in the legislation, that we are not looking at AI in terms of protection and the strongest protections we can possibly put in. We are looking at it in terms of the fear of what would happen if we put in those strong protections, whether those companies would stay and the implications for the tax take, jobs and all of those issues. That sort of colours the lens through which we are looking at this legislation. There is a fear from Government. We have seen it. We have talked about Grok. Within the space of a week of Grok being known, there were 200 reports to An Garda Síochána of child abuse cases and women as well, yet we are chasing our tails with all of this. We are chasing our tails because the legislation is not strong enough primarily because we have eight of the largest companies driving this. How strong is the witnesses' fear in terms of the protections for children? How deep are those concerns? I have them.

Mr. Timon talked about small indigenous business. I agree with him that over-regulation can sometimes stifle business and stifle the ability of small, indigenous businesses. However, I worry that the small, indigenous businesses will be buying these packs from these large companies. I am interested in his views on the connection between those who own all of this and are driving this, and small, indigenous Irish businesses. I do not think it is regulation or the cost of it. The ownership of this will be the big issue for Irish indigenous businesses. Are we big enough? Do we have big enough companies to have the power to override the big American companies we have here?

Ms Noeline Blackwell

Deputy Donnelly is 100% right. The development has been phenomenal in recent times. The rate at which it moves means that it is not only about this enterprise-based committee. The media committee had a report out about online safety this morning in that context and there is the AI committee. All of the Members in these Houses understand how important it is as an issue.

Regarding this particular Bill, we were so troubled about how to restrict it that we decided to focus on a couple of things that it might be possible to do no matter what the circumstances are. The EU AI Act twice in its recitals specifically identifies children as a vulnerable group, but nothing flows from it to the extent that even deepfakes are not considered prohibited. The European Parliament and Council now seem to be agreeing on that, but nothing will come into place until the end of this year at the soonest. There is a real issue around the protection of human rights in relation to AI in general at EU level. That is why we have said what we can do in Ireland in relation to any size of company that is producing harmful material or not taking enough care that it is not harmful.

If this piece of legislation is likely to be subjected to a regulatory impact assessment anyway, because I believe most legislation has to be although I do not see them that often, a children's rights impact assessment could also be done to identify where the harms are and to ensure they are plugged or offset.

If we have one huge concern, it is what people do when harm is done or when they are trying to protect themselves from harm. This legislation does not provide an answer. It does not say how the various fundamental rights agencies will be there for the people of Ireland to protect them either in advance through proper risk assessment or where they go when they have a complaint. That is going to be a real issue in terms of compliance.

Mr. Victor Timon

We know some of the large systems are owned by some of the big tech companies and it is right that there is a small number of them. We cannot tell people to get rid of their Microsoft Word system because it is owned by a big American multinational. A lot of this is about the data of companies. They can use these big systems while protecting their own data. From a protection point of view, as a lawyer this is what I encourage people to do. We are trying to build a system where these people are now able to compete on a global scale by the use of these products and protect their own position and data at the same time.

Senator Linda Nelson Murray

I thank all of the witnesses for coming before the committee and giving very insightful presentations and opening statements. We are all learning so much every week about AI. Usually at this committee I jump straight onto small businesses but I will park that and come back to it. I want to speak to the Children's Rights Alliance representatives. I have two girls aged ten and 11, whose minds are beautifully innocent. We do not allow smartphones in our house. They are not allowed on YouTube apart from timed kids' YouTube. They can use tablets for four hours at the weekend but only to game. I am so lucky that I am still in this bubble but in about a year and a half, when my older daughter will be 12 and a half and going on 13, she will get a phone going into secondary school like everybody else does, unless something mad happens. She is going to go from being that innocent who does not see anything of this world and, as has been said, we are just so scared of what they can see. To think that in the general scheme of the AI Bill we do not have specific lines when it comes to children is unbelievable.

I agree that children are vulnerable but it is not even that simple. Children interact with technology completely differently from us. If I were to ask one of my little girls to do something with a photo, it would take me an hour but she would have it done in a couple of minutes. They interact so differently because they are brought up with computer-type devices. They require explicit recognition and enhanced protections when it comes to legislation. Children are some of the most exposed users of algorithm-driven technologies in social media when it comes to educational tools and AI-generated content. There is a significant gap that we absolutely need to address. We really need to address this when we are writing our report.

As Deputy Donnelly said, some of the major tech companies are based in Ireland. Rather than looking at it negatively, we have an unbelievable opportunity to be ahead internationally by dealing with these companies and making sure that protections and AI governance is there. I am getting emotional about this. I am stopping myself tearing up.

Ms Noeline Blackwell

I thank Senator Nelson Murray. It is an emotional topic, particularly with regard to children, and it is an important topic. We spent our morning at Leinster House because earlier the Oireachtas Joint Committee on Arts, Media, Communications, Culture and Sport, chaired by Deputy Kelly, came out with something on online safety in general, about what we should do with this thing, which is not going to go away, and how it should be managed. It makes recommendations on algorithms. I would put AI into a slightly different category. I would say in relation to all of these issues that if the large media companies Deputy Donnelly referred to are putting out products that are not safe for children, why do we not require that they are safe in the same way we require a car to be safe?

Senator Linda Nelson Murray

A car seat for a child needs to be safe.

Ms Noeline Blackwell

Yes, as does a toy or baby formula. All of these things are required to be safe. That is doable under a different system of regulation. What is happening here with AI is that it has created a new tool in this generative AI, which is driving the potential for existing harms through faster systems with more sophistication all the time. Since we wrote our submission on 13 April, it has become even more sophisticated. The need is to make sure there is something to meet this new tool that has a greater potential for harm.

We are saying measures can be taken. We can make sure fundamental rights agencies, such as Irish Human Rights and Equality Commission, are properly resourced and understand what to do with this when something happens and, as Mr. Timon was saying, that there is co-ordination between all of these organisations. We can put in place children's rights impact assessments and the AI office could have a child, perhaps Senator Nelson Murray's child, on the board. These are the things that will deal with each tool as it comes along in a general way.

Senator Linda Nelson Murray

We run an SME test when we look at legislation that will impact small and medium-sized businesses. We absolutely need to be doing the same when it comes to children.

Ms Noeline Blackwell

Exactly.

Senator Linda Nelson Murray

From Digital Business Ireland's perspective, how do we ensure the legislation becomes an enabler of innovation rather than simply another regulatory burden for SMEs? Does Mr. Timon believe the general scheme properly reflects the reality of micro-businesses in Ireland where the owner, such as in my business, might be doing the payroll, the wages, the IT, the marketing and the rostering? How do they equip themselves properly for AI? Supports and grants were mentioned. What specifically would Mr. Timon recommend?

Mr. Victor Timon

On the supports and grants, we have recommended that the digital growth support be increased from €5,000 to €35,000 to get people to the next step. This would allow them to invest in something meaningful, such as a data analytics programme, and it would not be only for tiny or small licences. We see with the take-up of the voucher that it is not enough for people. We have also suggested, similar to green products, that there would be a tax allowance in the first year so people could use the allowance against their digital investment rather than spread it over the eight years.

In terms of supporting small businesses, we want to see more positive action. We understand why all of the regulatory and enforcement measures are there but we want to see the office mandated to take some positive action to help support businesses with education, compliance and support. Examples would be helping people with how they comply with something, explaining what legislation means, upskilling workers who may get displaced by AI and teaching others to become familiar with it.

An Cathaoirleach

The witnesses are all very welcome. This is a really interesting topic of discussion. Each of us is making sure the committee does its level best to give this the scrutiny, dialogue and discussion that are required. One of the issues that have been raised is AI content and the impact it is having on our social media feeds. We have heard from committee members about the impact it can have on children and about ensuring safeguards are put in place. The line Department we are dealing with on this is the Department of Enterprise, Tourism and Employment. My first questions are for the Children's Rights Alliance representatives. Are they concerned about this being discussed through the prism of enterprise and commerce? Does they feel they are getting sufficient attention from the Department in respect of this?

Ms Noeline Blackwell

I will defer to IHREC in a second. We had a small discussion about this earlier.

However, it is bound to happen. To pick up on what Deputy Donnelly said earlier, we see innovation the whole time and it is important that there is more recognition of safety and human rights protection in that area now. We cannot get away from it. It is where we are, but what we would love to see an equal emphasis, every time innovation is spoken about on recognition of the need to preserve the human rights of the Irish people.

Mr. Liam Herrick

It is a fair point. We recognise that several committees of the Oireachtas have taken an interest in this issue. The fact the line Department is one with a remit for business is a particular challenge. We note that we agree very much with the submission of Mr. Timon about the need for co-ordinated regulation. It is a complex regulatory system and co-ordination is essential.

However, we note that so far quite a bit of attention and consideration of resources has gone into the bodies designated as market surveillance authorities but almost none has gone into fundamental rights bodies. For example, IHREC is one of the nine named bodies but no consideration is given in the general scheme to any dedicated resourcing for us to carry out that function or to how we will co-ordinate with the other bodies and that is a particular challenge.

Some of this comes back to the question of the independence of the AI office. We are supportive of the AI office having a function in innovation and it is provided for in the general scheme. However, as Deputy Donnelly set out, as there is a strong commercial interest here - we all recognise that and it is nothing to be afraid of - there is all the more reason for strong and robust independent regulation. We have been down this road before with regard to the regulation on data protection and Coimisiún na Meán about the need for robust, well-resourced, independent regulatory bodies. The approach taken to date has been quite weak in that regard. That is how we would characterise it.

An Cathaoirleach

I appreciate that information. I just wanted to open my line of questioning.

The second side of it is for Digital Business Ireland. In his opening statement, Mr. Timon referenced that innovation adoption by AI businesses in Ireland could be done better. In the past seven days, I had the opportunity to visit IBEC to participate in a discussion with Digital Europe. One insight I found interesting was a key finding from that meeting, which is that the multinational sector - international companies and large-scale employers - is adopting AI at a pace five times faster than SMEs in Europe. I find that quite disturbing because they have such a competitive advantage. Does Digital Business Ireland have any particular message on those two matters?

Mr. Victor Timon

The larger companies have the resources, including money, research and people, so they will always be able to do that faster. On the take-up in Ireland, last year 20% of companies said they used AI. We are below the European average for that for SMEs. Countries like Denmark and Finland are at almost 40% so the take-up is low. Will a small indigenous company in the west ever compete with a global multinational? We do not know but AI is an attempt to level the playing field in what is achievable. Some of the legacy systems that some of those companies have spent a fortune building will be replaced quickly and cheaply by AI. There are great opportunities and all we are saying is that we should make sure SMEs get those opportunities and that the AI office should be mandated to do more than just the very good things my fellow witnesses need. It should be mandated to help business to grow.

An Cathaoirleach

Absolutely. Does Digital Business Ireland have any proposals for what that could look like? Should it be a team in the Department of enterprise or Enterprise Ireland working with companies that want to upskill?

Mr. Victor Timon

An example we looked at was the Sustainable Energy Authority of Ireland. It provides centralised guidance on regulation compliance. It runs seminars and tutorials. It puts interested parties in touch with experts in learning skills and such issues. We have met separately with such people as the Data Protection Commission. We always have the same message, which is that regulation is great and important but agencies should try to help people so they do not have to penalise them. They should try to help people to get to a place where they are doing it properly and they do not have to be fined. We want to see a more positive attitude, as well as the things my friends here need.

An Cathaoirleach

I appreciate that.

Deputy Brian Brennan

I thank the witnesses for their time. I am sorry I was not present for the opening statements. However, I read them. AI is in committees more often than I am and I am in them quite a lot. It is such an evolving situation with AI that sometimes I think it is like a raging bull and we have to try to catch up. It is not a lamb; it is a bull. We are trying to keep up. I know for a fact that if the people in this room sit down again next year, we will be having a completely different conversation. That is what we all have to take in hand. All we can control, as was said in our AI meeting this morning, is the controllables. We can try to deal with what is in front of us. The most important reason everyone here was elected was to protect our people and that is what we have to do. That is our number one job in these Houses. That is where I am coming from on this.

In reference to that, last night I wrote down the term "AI police". If there is a complaint, who will deal with it? While I could be wrong, when reading the opening statements, I detected more than a little concern about that in terms of the AI head office and how it will operate. Who is accountable if they are found to be wrong? Penalties have been outlined, but we have seen it previously when we are dealing with these multinational companies that were spoken about earlier. I heard it in my office before I came down. Who is policing these people? I would be interested to hear whether the witnesses are concerned about how the whole sector will be compliant? I am interested in opinions across the panel.

Ms Rebecca Keatinge

Our overriding concern is how this will work in practice and how it will be effective and coherent because we are part of the regulatory architecture, as the chief commissioner set out. There are 15 market surveillance authorities and nine fundamental rights bodies. As the general scheme stands, it is not clear that there is a market surveillance authority for each of the high-risk areas, for example, within our scope. There is a gap already there. I understand that when Department officials were before this committee recently, they indicated there may be further designations. However, for example, with regard to essential public services, which fall under the high-risk category in certain instances, it is not clear who the complaints body will be.

Second, the general scheme refers to individuals being able to make complaints to each of those market surveillance authorities under existing complaints mechanisms. Each of those bodies has a different mechanism. Even the fundamental rights authorities all have different statutory bases, different functions and different powers, so there is a lack of coherence. There needs to be a rethink of how this will operate in practice. Some of it will need to be on a statutory basis. In our written submission, IHREC highlighted where we see gaps and how this could be tightened up but it is practical and sensible to look at it from a layperson perspective. If people have an issue, how will they canvas and progress it?

The final thing I will mention, which we also suggested in our submission, is transparency measures. As a layperson, it is difficult to know whether there is a use of AI in the decision of a public body, for example, affecting a person's right to an entitlement. We have seen in other jurisdictions how it has been in that space that very vulnerable people have been affected, sometimes in devastating ways, by the use of AI. Therefore, IHREC has suggested that a national public register be put on a statutory basis.

There is provision for a register within the general scheme, but that is on a very narrow basis following the AI Act. I just flag that as a further consideration.

Deputy Brian Brennan

Okay. Does Ms Blackwell want to come in on that?

Ms Noeline Blackwell

Yes. I thank the Deputy very much indeed. From our point of view, the Children's Rights Alliance is a non-governmental agency, but it is somewhere people can go and say they are being harmed and ask where they can go to complain or who to talk to. When we look at this conglomerate of organisations Ms Keatinge outlined, we do not know where we will be able to send people to pursue their complaint. We do not know how we will help them to work up that complaint so that they know what they are looking for. Genuinely, we do not know. Do we send them to whoever it is? There is nothing about that, so that co-ordination of the nine fundamental rights agencies, which all do different things, is important. Some of them are regulators. Who will we refer to what? That is when the harm has been done.

Before the harm, we are really concerned that this is not being thought through from a rights perspective. That is where we say we should put a lens over everything we are doing here. We should put a human rights and children's rights lens over it and ask what the impact is. Are we taking the risks into account on what is happening here? This is already something Government policy states that children's rights impact assessments should be used for. This is something that could be done in advance to prevent the harm. Finally, we say put children in the AI office or on the AI office board, or child advocates if people do not want to have to work with children - we should really have children but we will settle for the other - and ensure that the legislation has a specific requirement whereby the human rights interests of children are represented on that board because the way it is right now is far too casual.

Deputy Brian Brennan

Okay. I would like to get a comment from Mr. Timon on that. With his legal background, are we not producing a Bill here that we do not know how to regulate?

Mr. Victor Timon

How to regulate AI?

Deputy Brian Brennan

No. To regulate the compliancy of the Bill.

Mr. Victor Timon

My friends here have eloquently put forward human rights and the interests of children whereas I have really only focused on the business side. They probably have similar examples but my concern would be if something happened in fintech through AI and there was a data breach, who is the primary person responsible? Do I go to the Central Bank of Ireland, CBI, or the Data Protection Commission, DPC? These are the kinds of issues we are concerned about. It is not clear what happens when there is a conflict.

Deputy Brian Brennan

That is exactly the point I am making. We are putting all this effort into creating this Bill, but it cannot be policed. I am not saying it cannot be policed, but it is not laid out very clearly and that should be the first part of a Bill. There is no point in putting this up as a front-page story. We need to be able to get underneath it and carry it out. If there are actions that need to be taken, who do we go to, what are the consequences and who hands out the consequences?

Deputy Sinéad Gibney

I thank all the witnesses for their contributions today. I will start with the overall governance principles that inform the establishment of the AI office. I would like to direct my questions, first of all, to IHREC because it is an established independent agency, arguably the most independent in the State, as it is governed by the Paris Principles. I have major concerns over the establishment of this office being in the Department of enterprise and remaining under the aegis of the Department of enterprise. I realise that it will become an independent State agency. Similar to the pathway we saw with Coimisiún na Meán, however, when an agency originates in a particular Department, I believe it is imbued with the principles and competence of that Department, which is through an enterprise lens. Even though it is with the Department of Enterprise, Tourism and Employment now, enterprise is what I believe is informing this general scheme.

My question is for Mr. Herrick. I know he touched on this within his submission, but what does he believe would be a stronger basis for the establishment of this office? He mentioned it is challenging that this is done through the Department of enterprise. Does he believe the Paris Principles should be applied here? He recently made a recommendation, for example, in relation to the International Protection Bill, which referenced the Paris Principles, but I do not believe that he made similar recommendations with regard to this general scheme. Perhaps he could outline why.

Mr. Liam Herrick

I will begin and Ms Keatinge might add to it. First, we are very conscious in this regard that the EU AI Act is directly effective and has direct effect in Ireland. In implementing legislation, there is a limit in the discretion that is available to the State. What we have said is that we have made a number of recommendations in terms of the Bill, as it has been presented, on how the independence could be greatly enhanced, although not going as far as the Deputy suggests with regard to the Paris Principles, which apply to national human rights institutions. These are practical steps that we think within the chosen approach could greatly strengthen it. We could, for example, remove the power under head 10A of the proposed general scheme, which allows the Minister to issue binding directions. That would be very significant. Even if that were to be kept, for example, it is understandable that any agency set up by the State should need to be in compliance with certain governance and regulatory requirements, we could exclude that the Minister could not give any direction with regard to individual cases, to investigation of enforcement decisions, sanctions or interpretation of the Act. We could be very clear that even if the Minister is making direction, it is on governance issues and not on substantive matters. We could also, for example, make clear that any directions would have to be consistent with the EU AI Act so that it limits it within that.

We have also suggested that the Public Appointments Service could be brought in with regard to the appointment of board members, as happens in a huge range of public bodies. That would greatly strengthen independence as it is perceived and enjoyed in practice. A very standard provision would be protections around the independence of the chief executive officer in terms of their protection from removal and also that they cannot be directed to take certain actions. What is of particular concern here is the proposal that the chief executive officer, when appearing before an Oireachtas committee, for example, would be curtailed and restricted in what they could say or comment on. There are very standard practical steps that the committee would be very familiar with in other contexts. There is a spectrum of independence in that regard and we could move it quite a bit on that spectrum.

Deputy Sinéad Gibney

Can I just throw something else in there? For me, there are also conflicts that are presented whereby, for example, we saw this week and last week that 700 Covalen staff are to be laid off because of the deployment of AI within the company. These are AI agents that, essentially, Covalen staff trained that are now taking their jobs. In the same week, in China, we saw two pieces of legislation that will now prevent companies from laying people off because of AI replacing their jobs. I have concerns that being born of the Department of enterprise will not facilitate effective enough consideration, for example, for the labour displacement. Indeed, it presents a conflict whereby the Department of enterprise supports businesses in being lean, mean and efficient but it does not do the same in relation to employees. Does Mr. Herrick see that conflict appearing?

Mr. Liam Herrick

The Deputy is absolutely correct. When we look at some of the functions that are set out in the general scheme, we do not have any difficulty with the AI office being competent with regard to promoting digital literacy. That is perfectly sensible because that is a benefit to everybody. However, the Department itself obviously has a primary responsibility to support the development of industry in Ireland. We also have the Industrial Development Authority, IDA, Enterprise Ireland and a huge range of other bodies. There is a real problem with the AI office also having this role to promote the expansion of industry and, at the same, time having an oversight and regulatory role in that regard. It is not to say that the Department should not be doing it. Of course it should, but there is a problem for a body that is quasi-independent having that function. We have been down this road before about how we ensure that we create the appropriate architecture in such a sensitive area of public policy.

Deputy Sinéad Gibney

Ms Keatinge may wish to add something. I will ask about the digital omnibus in the final minute and a half I have. The witnesses touched on that as well. I have raised it in multiple committees. I raise it nearly every day now. Indeed, I got it into the committee report. It was not quite in the printed version that was issued this morning, but it is in the online version that the Government consider its position on the digital omnibus, which I believe will roll back rights both in the digital omnibus package for general data protection regulation, GDPR, and our rights and protections under that legislation but also now under the AI Act before it is fully implemented in August of this year. Does Ms Keatinge have similar concerns? She has outlined specifics around the digital omnibus but maybe she could summarise her position.

Ms Rebecca Keatinge

It is very fresh. It was just last Thursday that it was finally got over the line and we are waiting for formal ratification, as it were. Our main concern is around the timeframes. We have this lacuna, essentially, between now and the end of 2027 where-----

Deputy Sinéad Gibney

I am sorry to interrupt, but does Ms Keatinge believe the self-regulation piece is-----

Ms Rebecca Keatinge

I think the Deputy is referring to the exception under Article 6(3). As I understand it, that is not being taken away. It has been held. That is a positive. Our overall view is that there is a concern about that gap in the timeframe where high-risk AI will essentially be unregulated for a lengthy period. Any high-risk AI that is currently in use will only have to comply if a substantial modification is made to it after December 2027. That is a concern. Some of the things we were concerned about with specific regard to our own functions were not implemented. That is a positive. Critically, we feel the risk associated with the exemption under Article 6(3) and the requirement to register if designating as non-high-risk has been averted. From our perspective, it is not all bad. It is part of an overarching simplification agenda. These tools are meant to be for technical amendments and not for wholesale changes to really important regulatory architecture.

Mr. Liam Herrick

An interesting innovation in the recent negotiations between the Parliament and the Commission was the introduction of a new provision prohibiting AI practices around the generation of non-consensual sexual and intimate content. I commend Michael McNamara MEP, who was one of the rapporteurs, for bringing that forward. There have been some positives but the overall direction towards deregulation and this big gap when there will be the possibility to introduce technologies with oversight are the overriding concerns.

Deputy Brian Brennan

I will pick up on a couple of points from the opening statements. Ms Blackwell mentioned that she was concerned at the lack of mention of children. The most vulnerable are mentioned. I was brought up with the idea that the most vulnerable in society are our children. Should I be reading something into her comment?

Ms Noeline Blackwell

The EU AI Act, the big parent Act, specifically notes that you have to recognise the vulnerability of children, not because they are particularly vulnerable in their own right but because of their stage of development and so on. This comes across in two recitals but there is very little follow-up in that particular area. When it comes to the general scheme of the AI Bill, which this committee is dealing with, even that does not come through with regard to children. If children are particularly vulnerable, which they are, and if we have online safety codes and a Digital Services Act that require special regard for what are called "minors" in the Act, the under-18s, that should follow through into other legislation. If it is not there, it is an enormous gap because AI can be particularly difficult for children in areas other committees will be dealing with, such as the area of child sexual abuse material online. Children are vulnerable to misinformation and disinformation in that they may not understand that what they are looking at is AI-generated. Not only should there be specific mention, but there should be a specific lens attached in order to view things from the perspective of children's rights in the implementation of this legislation and measures taken thereunder.

Deputy Brian Brennan

In relation to small business, people coming in have worn a track through that doorway. I am concerned about the levy and its effects. Small businesses in particular are already being hammered with costs. Is there any idea of what the levy will be? With regard to business in general and Deputy Gibney's earlier comments, what is the gut feeling out there regarding AI replacing existing jobs?

Mr. Victor Timon

On the levy, we have no idea what the figure will be. We are saying that it should be proportionate. We hear about the increased costs to keep small businesses going every day. We do not want any further burden on them. As I mentioned earlier, there is an advisory committee on reducing costs for small businesses. It would be counterintuitive to then impose more costs. If those levies are brought in, they should be set at a very low level and be very proportionate.

Deputy Brian Brennan

Will Mr. Timon address the displacement of jobs?

Mr. Victor Timon

The recent ESRI report states that there will be some displacement of jobs. We all recognise that. I see it in my own profession. They actually seem to be at the higher end rather than the lower end, which is where new technologies have traditionally replaced jobs. However, there are also great opportunities. We need to focus on the opportunities to create jobs. That means upskilling people who may never have dealt with it before to deal with a new world of AI. One of the things we are saying about this office is that it should be looking at this and that one of its remits should be the education of people to deal with and work with AI.

Deputy Paul Donnelly

A couple of issues really concern me. Let us take the trade agreement with Israel and the fact that, despite huge human rights abuses, genocide and all so on, there has been virtually no enforcement of what was written in that trade agreement between the EU and Israel. Such conditions are not even included in what I am looking at here. There are no checks and balances or details on enforcement, including who is to carry out that enforcement. I will not mention a specific country but what would happen if two or three countries decided not to enforce this because huge companies are paying them huge amounts of tax? That should be considered.

IHREC has recommended that the general scheme be amended to include provisions requiring public bodies to register all AI systems used in the delivery of public services. What does that actually mean?

On the levy, who is to be levied? Is it proposed that all companies would be levied or just those that use AI? I know those were three big questions.

Mr. Victor Timon

To very quickly answer the last question, it is companies that use AI systems.

Ms Rebecca Keatinge

On the register, it was actually a recommendation of the Oireachtas Joint Committee on Artificial Intelligence to include a transparency mechanism. It is just one of many different transparency mechanisms that could be instituted. This is an opportunity to empower citizens to know when AI is in use in the provision of public services, which we see as a critical risk area. We have set out some references to other jurisdictions where this is in place. I was looking at the Scottish model this morning. There is a website setting out the register. It is designed to provide both general information and information for people who are familiar with the technology. People can follow the journey of the use of AI. It is also in use in France and in the Netherlands. In the Netherlands, it was a reaction to a number of scandals involving discriminatory algorithms used by public bodies. As yet, it is not mandatory. That is what we are recommending. That is just to provide more detail.

Ms Noeline Blackwell

I will come in on the enforcement. The Deputy is right; there is the law and then there is the enforcement of it. Key to enforcement if there is political pushback against it is providing ways for individuals to complain, to make their point, to follow through and to get a remedy. This legislation entirely lacks a method for a child or their caregiver, parent or guardian to complain where their rights are violated by the use of AI. There is no adequate system for them to direct a complaint to somewhere where they can get an adequate remedy. In general, the whole digital system lacks sufficient enforcement and an adequate complaints system. We will keep making that point because, sooner or later, someone has to bring it in. It would be of great value if this legislation included a co-ordinated system for situations where we needed to refer something to the Data Commissioner, IHREC or the Ombudsman for Children. All of these are recognised bodies but, if somebody is in trouble, you cannot tell them to go to all of these nine agencies to work out which might be right. Those bodies may not know themselves.

I am not denigrating the Irish Human Rights and Equality Commission, but it does not know how it is to work.

Mr. Liam Herrick

The AI Act provides that there will be fundamental rights bodies that have the role of standing up and providing a safeguard. However, the intention of the Government is that there will be no resources for that function to be carried out, there will be no technical support for us or the other eight bodies to carry out that function, and there is no clear provision for how we will link in with the other bodies that are provided for. It either means something or it does not. It is clear that thought has been given to how to resource and support the market surveillance authorities, but not with regard to the fundamental rights bodies. The potential is there; a framework is there. It requires the Government to support the legislation with resources and commitment.

Deputy Sinéad Gibney

The potential impact of AI is only starting to come into focus for us, as a species. It is reaching into so many areas of our lives. It is powering and amplifying the algorithms that dominate our feeds and, in some cases, people's lives. AI chatbots are prompting people of all ages towards suicide. AI is being used in the area of defence. It is, as Ms Keatinge described, being used in places in the public sector where decisions are made in respect of vast swathes of our population. We cannot treat the issue seriously enough. We cannot examine it in too much detail in the interests of getting it right. As Ms Keatinge pointed out, the potential delays with the AI omnibus Act mean that AI will develop quickly in the 18 months in between and we will be left in the dust.

I want to put forward a hypothetical. If the general scheme of this legislation were now in effect and the architecture and infrastructure it describes were in effect, and it was in place when the Grok scandal happened, do the witnesses believe that there would have been any difference? The Grok scandal happened at the beginning of January. As far as I can see, we still have not had any prosecutions or recourse. I can still access the nudification app on my phone using a virtual private network and proliferate child sexual abuse material and image-based sexual abuse on the offering of an AI tool that is provided to me. I do not believe that we have seen adequate access to justice. Do the witnesses believe that if what is proposed in this general scheme had been in place when that scandal happened, it would have offered additional solutions?

Deputy Brian Brennan took the Chair.

Mr. Liam Herrick

I might answer that question in two parts. We have identified that the regulatory system in place at the moment is confused and inadequate. In that sense, what is provided for in the Irish legislation seems inadequate. On the other side, we now have a clear statement of intent from the Commission and the Parliament that there will be binding requirements in the AI Act that will prohibit the development of that technology. It seems there is now the promise of binding European legislation that will create legal obligations to prohibit usage, and that will have a direct effect.

The question then is whether it will prevent the technology being available or will just create a legal basis for recourse if people suffer as a consequence. Will it be preventative or will it just provide a remedy if a violation happens? We do not believe that the regulatory system here is adequate to meet emerging and new challenges, such as the challenge identified by the Deputy. On image-based harm, it seems there is the will at a European level to have binding regulations for that specific harm.

Deputy Sinéad Gibney

It is just whack-a-mole. Grok is a feature on an app on a platform. We do not know what the next Grok will be - it has not emerged yet - but there will be another scandal based on technology that we are not familiar with and that will cause untold harm. The suicide stories relating to chatbot use are incredibly concerning. I have raised the issue at meetings of the AI committee. At the moment, algorithms demand our attention in order that they can serve us ads to watch. Chatbots will, and do, create intimate relationships with us and reach into our most vulnerable emotions. An emotional tie is being created through these chatbot relationships, and it is hugely concerning to me that we are seeing this. It is under-reported, and more will come out. The next Grok is only around the corner.

An Leas-Chathaoirleach

Would anybody like to make any final comments? I have found this meeting hugely interesting. I hope, going forward, that we can take on board the witnesses' comments. It is such an evolving situation. As Deputy Gibney pointed out, what is going on and how quickly it is evolving is frightening. It is having an effect on every sector of our community, from kids to key businesspeople in the world. The policing of it seems to be ignored. Whatever issue you have, you should be able to go to somebody and ask how it can be dealt with. That seems to be totally missing at the moment. We need to act on that, and that is why we are here. Would anybody like to come in with any closing comments?

Mr. Liam Herrick

I will come in very briefly. We do not believe that innovation and the development of this new technology, and all the benefits it promises, are irreconcilable with a healthy system of regulatory safeguards and protections. We think both can be achieved. In that regard, we very much support the stated position of the Taoiseach and the Government. The commitment is for Ireland to be a leader in innovation with regard to AI but also to have respect for fundamental rights. It is particularly relevant to say in that context that we have recently written to the Minister of State, Deputy Niamh Smyth, and the Government has indicated that it intends to hold an AI summit as part of the Irish Presidency of the European Union. We have asked that alongside an appropriate emphasis on business and enterprise, there would also be a balancing focus on fundamental rights. We can continue this conversation and work ahead with development and innovation and a healthy protection of fundamental rights at the same time.

An Leas-Chathaoirleach

As a committee, we intend to take Mr. Herrick's comments forward.

Ms Noeline Blackwell

I thank the committee for hearing us on this topic. As Mr. Herrick said, Ireland's digital strategy is that it will lead on innovation with all due respect for fundamental rights and safety. That is so important and can be achieved.

I have from time to time heard Deputy Gibney say that this committee, the AI committee, the children's committee and the media committee are all bringing together various aspects of the same thing. Until there is a bringing together of the strands of the various committees, we will be allowing tech to take advantage of the siloed nature of each of these interventions. I hope that over time all the wisdom of the various committees can be brought together for a more comprehensive understanding of the AI tool, which will, as Mr. Timon said, provide an opportunity for great development and new jobs, but also presents a risk of real harm.

Deputy Albert Dolan

I apologise. I had to run out to a cardiovascular illness prevention talk. My questions, on the back of the opening statements, are around AI and safeguarding. It is already a hard question to answer because of the speed at which AI is moving. It is adapting and changing so quickly. The landscape is moving every day. How quickly some of the capabilities are improving is mid-blowing. A few weeks ago, Anthropic, which owns Claude, dropped a new design tool. What will that mean for graphic designers in their jobs? The quality of what can be done quickly is incredible. Where do we draw the line between legislation, State intervention and innovation? How do we have that conversation in a practical way? Ireland is a place with a huge tech hub and base. I want us to be a country of innovation, excitement and driving forward. Simultaneously, I want to make sure that our workers are protected, educated and upskilled to meet this opportunity. I believe that whichever country moves first and at speed with AI will see the greatest opportunities come out of it and the greatest return.

How do we draw the line? Where do we draw the line between State intervention and innovation? I would love to hear some opinions on that. Whoever wants to go first may do so. I really do appreciate the witnesses' time. It is something of a complex question.

Mr. Victor Timon

There is a place for both. There have been some very articulate made arguments today about the need for enforcement and views as to why, in some areas, it has not gone far enough.

In business, not just children's human rights, we need regulation. However, what we are pushing for is support for innovation, and I think the two can sit very comfortably together. We are not suggesting that some of the regulation should be got rid of because it is too regulation heavy; rather, we are saying the legislation should be enhanced with more business-positive functions, such as skills training and compliance support. A point we made to the Data Protection Commission previously is that within its remit, it is supposed to be educating people as to how to comply. It is much better to educate people on how to comply than to fine them when they do not. We would look for these types of supports where the role of the office is concerned.

Deputy Albert Dolan

On that point, the position is very interesting. Over the last few years, I have spoken to a lot of people about the National Training Fund and unlocking its potential. We know where it needs to be unlocked, particularly with the wave that is coming, but I appreciate Mr. Timon's remarks.

Mr. Victor Timon

We would say there is a lot of money in that fund. It seems an obvious place to start in terms of skills training.

Deputy Albert Dolan

Yes.

Ms Noeline Blackwell

The Deputy's question is tricky. We have been here before. Over decades, and even centuries, industries have had to introduce health and safety initiatives. Twenty years ago, we were all talking about how regulation of the finance industry was inadequate during the recession. We are now talking about the digital industry, and we want to do the best we can. When people look back on how Ireland and the EU in general have dealt with regulation of the digital industry, they may conclude we were asleep at the wheel for about ten years too long. There was development and we should have been saying that while we wanted the best possible products, they had to be safe, with safe systems, and respectful of rights.

We are not asking that anyone be slowed down, we are not asking that anyone be hampered in development and we are not asking that jobs go elsewhere; rather, we are asking that this country not be the one that takes the products while risking human rights, including children's rights, and safety.

Deputy Albert Dolan

I guess the fundamental point is that good legislation puts good guardrails in place within which the industry can thrive, develop and be of benefit to people.

Ms Noeline Blackwell

That is it.

An Leas-Chathaoirleach

I will allow an extra four minutes to give the witnesses time to answer, because we are discussing a very important subject.

Mr. Liam Herrick

The Deputy's initial question was on where we appropriately draw the line. We do not draw it by putting the central regulatory agency, the AI office, in, adjacent to or under the direction and control of the Minister for Enterprise, Tourism and Employment. It cannot be the independent arbiter if it is too closely associated with the State Department responsible for advancing innovation and industry and completely separate from the question of safeguards.

Deputy Albert Dolan

What body would we put it under? Should it be kept completely outside the Department?

Mr. Liam Herrick

We have made a number of practical suggestions. Even if the office's line Department or parent Department is the Department of Enterprise, Tourism and Employment, it can be placed at arm's length by using simple, practical tools that are used for statutory agencies all the time. These relate to safeguards concerning the power of the Minister to direct, the power of the Minister to appoint or remove, and the types of issues on which the agency can or cannot speak. There are practical suggestions in our observations on the legislation that would greatly assist regarding the office's independence.

I agree it is in everybody's interest that regulation provide certainty and clarity to industry, and that there should also be a well-thought-out scheme of co-ordination between the various bodies. Business benefits from certainty and clarity, and the public benefits from transparency regarding how the system is regulated.

Deputy Albert Dolan

To play devil's advocate, though not expressing an opinion, would it not make sense to have the AI office under the Minister for Enterprise, Tourism and Employment given that the area of enterprise and employment is likely to be one of the most significantly affected? Am I wrong to ask that?

Mr. Liam Herrick

We are accepting that the approach chosen is to give the office a line relationship with the Department, but that does not prevent significant protections being put in place to ensure it operates at arm's length.

Ms Rebecca Keatinge

I would add that the EU AI Act requires, under Article 70, that the office exercise its powers independently, impartially and without bias. To be consistent with that, and given that it has direct effect, the implementing measures need to be consistent with the Act. It follows from that.

Deputy Albert Dolan

The AI office is expected to have a relationship with other regulatory bodies, such as Coimisiún na Meán and others. I cannot think of all their names now. Are there 13 agencies the office is meant to deal with?

Ms Rebecca Keatinge

It is 15.

Deputy Albert Dolan

It is meant to interact with 15 agencies. Independence, both in appearance and reality, is vital if the office is to be the impartial adviser to all of those bodies, each of which will regulate its respective sector and have AI issues funnelled into it.

Ms Rebecca Keatinge

Yes, it has a complex role. It is being established to ensure consistent enforcement rather than to be an enforcer in and of itself. It needs to be able to interact with all of those agencies on an independent basis and without any direction. It will be responsible for making very substantive decisions about how that regulatory architecture will work.

Deputy Albert Dolan

For my benefit and that of the Leas-Chathaoirleach, if we were to take one message back to Government on this issue, what would it be?

Ms Rebecca Keatinge

Mr. Herrick has pointed to some of the amendments we have proposed regarding the office's independent functioning. However, one aspect that has not been amplified is the inclusion of fundamental rights as one of its functions. Innovation is a very important aspect but there is currently no obligation on the office to promote and protect the deployment of AI in a manner consistent with fundamental rights. This is a very important aspect. From that will flow the need to co-ordinate the bodies designated to form part of the architecture.

Deputy Albert Dolan

Could I be given an example of where AI could go against somebody's fundamental rights or might not be aligned with them?

Ms Rebecca Keatinge

There are a number of examples in other jurisdictions of where fundamental rights have been undermined through the use of AI. A very famous example is the child benefit system in the Netherlands. It had a fraud detection system that used a self-learning algorithm. Indicators such as citizenship and other factors were used to identify where there could be a risk of fraud. It picked on people who had made small errors in their applications and marked them out as a fraud risk. It had a highly disproportionate impact on ethnic minorities and people in the lower socio-economic bracket. It led to huge sums of money having to be paid back to the State. Twenty-six thousand individuals were affected. A compensation scheme followed, together with a number of retrospective measures, and it led to the downfall of the Dutch Government, as the Deputy may be aware. That was the turning point. It is a very large-scale example.

The AI Act contains a huge number of references to fundamental rights, particularly to the Charter of Fundamental Rights of the European Union, which covers huge swathes of rights. A number of equality directives are engaged. The equality standards directives are due to be transposed in the coming months and they will place new obligations on the State regarding the resourcing of equality bodies and enabling those bodies to access information to ensure there are no breaches of equality protections. I thought I would mention that because it is very timely.

Ms Noeline Blackwell

To add to that, erotic AI companions and chatbots are not regulated at all. There was a case in the UK in March in which a coroner made the point that a 16-year-old boy who took his own life the previous day had got specific instructions from an AI companion on how best to do so. There are many examples of where children do not distinguish between AI and reality. I do not always distinguish between them either, and I am no child.

It is just that there is such a risk of danger from these unregulated companions at all levels. Even the ones producing child sexual abuse material that are to be prohibited will not be prohibited until the end of this year at the soonest. As for other AI companions, they are not even seen as high risk at the moment. There are an awful lot of instances where people can be harmed.

I am sorry, a Leas-Chathaoirligh. I overshot by two minutes.

An Leas-Chathaoirleach

It was hugely interesting.

Deputy Paul Nicholas Gogarty

I listened to the opening statements. I missed some of the earlier part of the debate, so apologies if I am going over stuff. I would like to mainly focus on what people have in common and what they do not. For example, the Irish Human Rights and Equality Commission wants rights-based, child-based regulators and the Children's Rights Alliance is similar. Digital Business Ireland wants oversight as well but it wants it to protect business. I am trying to work on a common message from everyone. They all want to see some form of regulation but in different areas. From the business perspective, Digital Business Ireland does not want too much regulation because that would stymy business but it acknowledged that there still needed to be oversight, regulation and practical supports. Is there a win-win for everyone in having a clear set of regulatory rules? What do the Irish Human Rights and Equality Commission and the Children's Rights Alliance think would be a balance to protect business rights while also protecting children? To Digital Business Ireland, how can one encourage enterprise while ensuring vulnerable users, especially children, will not be abused by the system? In some ways, the groups are singing from the same hymn sheet and there will always be common good or sectoral interests, but there has to be a message from this committee to the Minister to inform the next phases of legislation of where there is common ground. I would be interested in the perspectives of the witnesses on that.

Mr. Liam Herrick

One can look at this from the question of quality of regulation as well. Where there is a common interest among all stakeholders, including business, is in having regulation that is clear, transparent, provides certainty, and is well structured and co-ordinated. That is in everybody's interest. Specifically, we think it would be important, and have made specific recommendations, that all agencies involved in the regulatory scheme have a responsibility to take into account fundamental rights. At present, that is not visible in all elements of the regulatory scheme.

There are specific bodies, including us, that are named as fundamental rights bodies. There is no provision for us to be resourced or have the technical capacity to carry out that function. We must ensure that every regulatory body that has an obligation to take into account fundamental rights principles and those that are specifically dedicated to that function are equipped to carry it out.

Mr. Victor Timon

We are very compatible in what we are looking for. All of us are looking for certainty. Good regulation of business, things like the prohibition of certain systems and the regulation of high-risk systems probably have a knock-on impact on protecting human rights and children's rights - perhaps not as much as Ms Blackwell and Mr. Herrick have set out, but the two sit comfortably together.

Deputy Paul Nicholas Gogarty

Does anyone else want to comment before I come back in?

Ms Noeline Blackwell

I echo what Mr. Herrick has said about certainty and recognising that regulation without understanding that fundamental human rights are engaged in that regulation is not going to work well. It does not protect the people of Ireland. That is where it is important.

Deputy Paul Nicholas Gogarty

I am being a bit of a devil's advocate. Does Ms Blackwell believe it is possible to have stringent rights but not overburden small businesses in particular?

Ms Noeline Blackwell

Yes. We would be in a much better place and much less worried about our entire populace, particularly our children, if we had started regulating social media ten years ago and had understood that one could not operate these massive businesses in an unregulated space without concern for safety and human rights. It is necessary that that be taken into account.

Mr. Liam Herrick

I agree. I think it is possible but one of the great concerns we have in terms of where we are now and certainly in light of what was agreed last week between the European Commission and the European Parliament is that it now seems that there will be a period up to the end of 2027 when the regulatory balanced system we all want and are trying to help create will not be in effect with regard to high-risk applications. That creates the appalling vista that there could be a period of time when products could be introduced without being subject to regulation. Given what everybody, including committee members, has said about the pace of development, that is a real concern. That has come through the so-called simplification agenda. That will be a challenge for Ireland and, indeed, the European Union more generally.

Deputy Paul Nicholas Gogarty

Our sister committee decided not to recommend a ban on social media for under-16s. It is something that I had suggested to the Tánaiste should be put in in the absence of proper meaningful legislation. He said that the era of self-regulation was over, but anything proposed so far by the Government appears to leave it up to companies to voluntarily enter and age verification checks have been left up to the general public to voluntarily engage in. Will the witnesses comment on whether this would be any different with AI? We need stringency but is it possible not to put the burden on business and for the regulatory aspect to be done from elsewhere?

Ms Noeline Blackwell

The report of the committee this morning, which did not recommend a social media ban, noted that none of those who had looked at this with a view to protecting thought it was a particularly good idea, not even the tech companies. They might say that, might they not, but neither is it effective from a human rights or children's rights point of view. The issue about regulation is it should not be burdensome but neither should it run a real risk of harm. That is the concern with the current availability of social media - it can allow our children and young people to be subjected to harm. It should be a safe product. Companies should and, we know, are able to produce safe products. It just might slow them down in terms of profitability or the amount of work they have to do. We expect safe products throughout our lives, particularly in the European Union. We expect safe financial products and health and safety in our workplaces. We are entitled to and should receive safe products and systems from the digital world, including those generated through AI.

Mr. Victor Timon

Given the number of market surveillance authorities and so on under the Act, one could not say it is a self-regulating market in Ireland. Even in respect of the omnibus, what had been mooted was that people would be able to self-regulate to say this was not a high-risk system, so they did not need to register it. That right has been taken away now as well. There is probably less scope for self-regulation.

Ms Alex Murphy

In relation to the proposed social media ban, while there obviously is significant discourse around that area, it is not really hitting the issues as Ms Blackwell said. What we are looking for are safe systems all around. For example, children are not only interacting with harms on social media. It is on gaming platforms. It is on most things that are digital in general and artificial intelligence is just exacerbating those already existing harms. If we looked at the ban from that perspective, it would be very difficult to impose an artificial intelligence ban for children considering it is being embedded across the country anyway. From the digital and environment perspective from social media platforms, all evidence is showing that children are just going to go to unregulated platforms where they will be in more danger and there will be more harm.

Deputy Paul Nicholas Gogarty

I thank everyone for their contributions. There is some common ground there and hopefully we can work that into the legislation.

An Leas-Chathaoirleach

I thank the Deputy. Do Ms Murphy or Mr. Timon wish to make a final remark?

Mr. Victor Timon

I am good. I thank the members for listening to all our comments.

An Leas-Chathaoirleach

They are much appreciated. I thank our witnesses for their contributions at today's meeting. I propose that the committee meets in private session on Tuesday, 19 May 2026 at 2.15 p.m. and in public session on Wednesday, 20 May 2026 at 12.30 p.m. Is that agreed? Agreed.

The joint committee adjourned at 12.31 p.m until 12.30 p.m. on Wednesday, 20 May 2026.

Joint Committee on Enterprise, Tourism and Employment díospóireacht - Wednesday, 13 May 2026

General Schemes of Regulation of Artificial Intelligence Bill 2026 and Data Bill 2025: Discussion (Resumed)

Mr. Liam Herrick

Ms Noeline Blackwell

Mr. Victor Timon

Mr. Victor Timon

Mr. Victor Timon

Mr. Victor Timon

Mr. Victor Timon

Mr. Victor Timon

Ms Noeline Blackwell

Mr. Victor Timon

Ms Noeline Blackwell

Ms Noeline Blackwell

Ms Noeline Blackwell

Mr. Victor Timon

Ms Noeline Blackwell

Mr. Liam Herrick

Mr. Victor Timon

Mr. Victor Timon

Ms Rebecca Keatinge

Ms Noeline Blackwell

Mr. Victor Timon

Mr. Victor Timon

Mr. Liam Herrick

Mr. Liam Herrick

Ms Rebecca Keatinge

Ms Rebecca Keatinge

Mr. Liam Herrick

Ms Noeline Blackwell

Mr. Victor Timon

Mr. Victor Timon

Mr. Victor Timon

Ms Rebecca Keatinge

Ms Noeline Blackwell

Mr. Liam Herrick

Mr. Liam Herrick

Mr. Liam Herrick

Ms Noeline Blackwell

Mr. Victor Timon

Mr. Victor Timon

Ms Noeline Blackwell

Ms Noeline Blackwell

Mr. Liam Herrick

Mr. Liam Herrick

Mr. Liam Herrick

Ms Rebecca Keatinge

Ms Rebecca Keatinge

Ms Rebecca Keatinge

Ms Rebecca Keatinge

Ms Rebecca Keatinge

Ms Noeline Blackwell

Mr. Liam Herrick

Mr. Victor Timon

Ms Noeline Blackwell

Ms Noeline Blackwell

Mr. Liam Herrick

Ms Noeline Blackwell

Mr. Victor Timon

Ms Alex Murphy

Mr. Victor Timon

Joint Committee on Enterprise, Tourism and Employment díospóireacht -
Wednesday, 13 May 2026