Oireachtas Joint and Select Committees
Tuesday, 19 May 2026
Joint Oireachtas Committee on Artificial Intelligence
Images, Deepfakes and Consent: Discussion
2:00 am
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I welcome our guests to this meeting of the Joint Committee on Artificial Intelligence. They are aware of the terms and conditions, as they say, of appearing before Oireachtas committees. Ms Blackwell in particular has been very engaged on these issues with us. We are focusing on questions we touched on previously in the context of issues around images, deepfakes and consent. This is a very important topic. I am glad we are returning to it. We are joined by Ms Noeline Blackwell, online safety co-ordinator, and Ms Alex Murphy, online safety legal officer, from the Children's Rights Alliance. We are also joined by Ms Olga Cronin, senior policy officer for the Irish Council for Civil Liberties. I invite Ms Blackwell to deliver her opening statement on behalf of the Children's Rights Alliance.
Ms Noeline Blackwell:
I thank the Cathaoirleach and the committee very much for inviting us to provide evidence for their consideration in respect of recent issues relating to artificial intelligence, AI, particularly the development of AI images and deepfakes and matters around consent. In it approach to Ireland’s forthcoming Presidency of the Council of the European Union, the Government is seeking to be an innovator and leader in the development of AI while at the same time endeavouring to ensure its safe development and deployment. This statement and the longer briefing paper we submitted to the committee focus on that safe deployment and development. We are gravely concerned about the current lack of regulation in respect of and the provision of safeguards for children in the rapidly developing technology of generative AI.
At EU level, the recent EU AI Act, which has not even been fully implemented yet, is being amended by the new AI digital omnibus in what aims to be a simplification of the existing legislation. While there can undoubtedly be merit in streamlining regulation, the current reforms mean that, at best, whatever limited safety net might be put in place in relation to images and deepfakes will be further delayed. As we understand it, regulation of high-risk AI applications that should have come into law this year will not now be in place until the end of 2027 at the earliest.
Indeed, even if the law were not under review, many of the risky AI chatbots that impact children would not have been recognised as high risk, partly because of the speed of development of that technology, but also because children’s rights and their usage of technology were not, and are not, properly considered in the EU legislation. Even the recent welcome announcement that the EU Council and Parliament have provisionally agreed to prohibit AI-produced child sexual abuse material, CSAM, is not enough to protect children now, with the likely implementation date for that being either the end of 2026 or early 2027. Indeed, it is a sign of the lack of understanding of children’s rights and welfare that it has taken until now to recognise that such AI productions should be prohibited.
Deepfakes pose greater risks for children than adults because children's evolving capacities and development mean they have more difficulty identifying deepfakes. Children are also more susceptible to harmful online practices, including grooming, cyberbullying and CSAM. As was stated in the Chairman's press release yesterday, according to UNICEF, at least 1.2 million children disclosed that their images had been manipulated into sexually explicit deepfakes in the previous year according to its 2025 study. The non-consensual sharing of deepfakes, and very often even the apparent consensual sharing, constitutes a fundamental assault on a child’s dignity and bodily integrity. They are not just illegal; they are truly harmful. As a result of their generation, publication and dissemination, these images can be used to blackmail, bully, groom, threaten and abuse children and young people. They violate children’s rights to protection from serious harm, abuse and degrading treatment, their rights to health and safety, and their rights to privacy. Apart from the AI omnibus legislation, the current slow progress of a recast directive and a regulation on CSAM at European level impairs detection and prosecution of CSAM and fails to hold the tech industry accountable for its role in disseminating this material.
While recognising the need to deal with issues at an EU level, there is also a role for Irish legislators. While Irish legislation, especially the Child Trafficking and Pornography Act 1988 and the Harassment, Harmful Communications and Related Offences Act 2020, known as Coco’s Law, has capacity to prosecute some offences, we believe there may be gaps such as the generation of AI deepfakes and the tools associated. A current audit of Irish law being undertaken in the context of the EU recast directive may provide evidence for other legislation that could be undertaken to protect children at Irish level.
Still affecting children and young people, the ever-increasing use of AI in school programmes is a concern, particularly to children’s rights to privacy and data protection. Currently, there are a wide variety of educational technology, or edtech, applications and programmes used in Irish schools which have artificial intelligence models embedded. While these programmes may be beneficial to children in a wide number of ways and it is important and necessary that artificial intelligence is used to its best effect to empower children and young people, it is equally necessary that those applications used in schools are well understood, and that they fully respect all the rights of children and young people. Protecting children’s privacy means ensuring that their data or content about them are used only with their informed consent and legally. Currently, no system exists for the certification, assessment or licensing of edtech systems in schools. As a result, there is no guide for educators, nor is it possible to assess the extent to which the various current programmes comply with children’s rights in this regard.
There are recommendations attached to our submission which I am happy to share with the committee at a later stage if needed.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I thank Ms Blackwell. We can explore the recommendations through members' questions. I invite Olga Cronin from the Irish Council for Civil Liberties to deliver her opening statement.
Ms Olga Cronin:
The Irish Council for Civil Liberties, ICCL, thanks the committee for this opportunity to discuss artificial intelligence and the recent issues around images, deepfakes and consent.
On January 6, together with Digital Rights Ireland, ICCL wrote to the Garda Commissioner calling on An Garda Síochána to urgently investigate X Internet Unlimited Company, with an address in Dublin, under the Child Trafficking and Pornography Act 1998, for its Grok AI chatbot generating non-consensual deepfake sexualised images of children. It followed widespread media coverage of X’s Grok chatbot responding to user prompts to remove clothing from images of children and posting them on X. Reuters described it as a "mass digital undressing spree".
We pointed out that under section 5 of the Child Trafficking and Pornography Act 1998, it is illegal to knowingly produce, distribute, print or publish any child pornography, or to encourage or knowingly facilitate such production, distribution, printing or publication of child pornography, which is an unfortunate legal term.
The definition of child pornography includes any visual representation or description of a child which indicates or implies that the child is available to be used for the purpose of sexual exploitation. We also pointed out that under section 9 of the 1998 Act, a director or other officer of a body committing such an offence may also be guilty if the offence was committed with the consent or connivance of that person or is attributable to any neglect on the part of that person. We therefore asked that An Garda Síochána also investigate the culpability of the officers of X.
Given the gravity of the industrial-scale illegal activity and the great level of distress and public disquiet it was causing, we said that we believed An Garda Síochána had to exercise its powers urgently to address this issue. We also said that it was critical for An Garda Síochána to reassure the public that this deeply disturbing crime was being investigated and prosecuted. A week later, on 14 January, as the scandal continued and amid Cabinet confusion about what could be done, we published a table of relevant laws outlining the investigatory powers and tools that could be used by, An Garda Síochána, Coimisiún na Meán, the EU Commission and the Data Protection Commission, DPC. We published this table of applicable law and shared it with all Oireachtas Members.
On 20 January, we wrote to the DPC to ask what, if any, steps it had taken in light of X’s Grok AI chatbot generating these images. Specifically, we urged the DPC to exercise its powers under section 134 of the Data Protection Act, which provides that the DPC can make a High Court application for the suspension or restriction of data processing where there is an urgent need to protect data subjects’ rights and freedoms. We believed that moment had already been reached.
In respect of An Garda Síochána, we subsequently learned, through hearings at this committee, that An Garda Síochána, by 3 March, had opened 244 investigations into images generated by Grok. Around half of those related to child sexual abuse material. We asked if any of the investigations were focused on Grok and the management of it specifically as opposed to the users creating the imagery. An Garda Síochána told the committee, “We will go wherever the evidence takes us.”
In respect of Coimisiún na Meán, we understand, again through this committee, that it worked with the European Commission on the matter. That ultimately led to the Commission announcing on 26 January that it had opened an investigation into X to assess whether the company properly assessed and mitigated risks associated with the deployment of Grok’s functionalities into X in the EU. The Commission did so in the context of its powers under the Digital Services Act, DSA. Coimisiún na Meán was later asked by a member of this committee why it did not launch its own investigation under Ireland’s online safety code to run in parallel with that of the European Commission. The member in question was told that it was the view of the coimisiún that the action being taken by the European Commission under the DSA was more appropriate. This committee was also told that it did not appear to the coimisiún that there had been a breach of the online safety code. At a sitting of the Joint Committee on Arts, Media, Communications, Culture and Sport, the coimisiún stated:
Had there been more video content, we may have decided to use the online safety code. The online safety code applies to video content and video-related content in relation to the measures within it. As it was predominantly images we were coming across, we considered that the most appropriate tool to have the most appropriate remedy was the Digital Services Act.
The Data Protection Commission announced on 17 February that it had opened an inquiry into X under section 110 of the Data Protection Act 2018, with the decision to commence the inquiry being notified to X the previous day. ICCL does not know the status of these respective investigations and we commend the committee for seeking answers. We believe Ireland has the necessary legislative tools and investigatory powers here.
Of course, the issues we are discussing here are not confined to Grok. We note, on foot of this scandal, the European Parliament's and Council’s recent agreement to ban AI systems that create child sexual abuse material, CSAM, or depict the intimate parts of an identifiable person, or them engaged in sexually explicit activities, without that person’s consent. This would apply to: the placing of AI systems on the EU market with the purpose of creating such content; the placing of them on the EU market without reasonable safety measures to prevent such creation; and deployers using these systems for the purpose of creating that content. The content in question can be images, video or audio, and companies will have until 2 December next to bring their systems into line. This is a welcome move, but the existing provisions in this area must also be enforced.
The Grok scandal demonstrates both the scale and the speed with which AI systems can facilitate the creation and dissemination of illegal and deeply harmful content. It also illustrates that Ireland and the European Union already possess significant legislative tools and investigatory powers to respond to such content, under our criminal law, online safety framework, data protection law and the Digital Services Act.
Public confidence depends not only on the existence of legal powers but on visible and timely enforcement.
We thank the committee and welcome any questions.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I thank Ms Cronin. We will go to questions and answers from each member of the committee. Each member has seven minutes for questions and answers.
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses. I will start with Ms Cronin. She said that she believes Ireland has the necessary legislative tools and investigatory powers. Has she any comment on the pace of the Garda investigation, given the industrial scale of the alleged illegality?
Ms Olga Cronin:
We are not privy to the information the Garda has. It has to carry out its investigations and gather evidence and act accordingly. For our part, when the scandal first broke there was a lot of swirling commentary about what could and could not be done. It also felt to us that the conversation went straight to what new laws we could bring in as opposed to looking at what powers and investigatory tools we have. That is why we took the action we took. We did write to the Garda Commissioner and urgently asked him to investigate because we were afraid or, at least, we did not know what was happening and we wanted to make that clear to the Garda so it could take action.
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Does Ms Cronin have anything to say about co-ordination? There is a number of actors here – An Garda Síochána, Coimisiún na Meán and the DPC. Is there an argument that there should be greater co-ordination?
Ms Olga Cronin:
If you look at all the different pieces of legislation, these digital pieces of legislation, in an ideal world they should work together like a jigsaw. There are different tools within them to tackle different things. For example, the DSA is looking at the systemic risk. The Data Protection Commission is looking at risks to the data protection rights of everyone. There was a real role there across Europe which is why a lot of European digital rights groups were so welcoming of the fact it has taken that action. Also, there is the Garda to respond to individual pieces of harm when people go to it. With Coimisiún na Meán, under the Online Safety and Media Regulation Act it has powers in respect of the online safety code in terms of video-sharing platform services. We would say it has obligations there and powers there. It has significant powers under the online safety code. There was a decision between Coimisiún na Meán and the Commission for the Commission to take the lead. I know a member of this committee questioned Coimisiún na Meán to see how come there was not a parallel investigation and, as I said in my opening statement, the answer was this is illegal content but if it had seen more illegal content it would have responded. That is a decision.
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
It raises the reasonable question of whether there is a specific threshold of illegal content that it needs to see before-----
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
No, I understand.
Ms Olga Cronin:
People in this room might remember that when the OSMR was being created there was a lot of questions about how this overlapped with the DSA and how the online safety code would act because there is overlap. There were many calls for a clear document outlining how these things would work with these different bodies. This was the first test of that.
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
From the point of view of victims and the public, is there any single public contact or joint update or statement to keep victims and the public abreast of developments in investigations? Are there supports available for victims in these cases?
Ms Olga Cronin:
I am not sure I am in a position to answer that. There are different regulatory bodies doing their own jobs. I have not seen any joint declaration to the members of the public but I am absolutely aware we have had the Garda and the hotline calling on people to individually contact them and I would absolutely echo that.
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I thank Ms Cronin. My next question is for Ms Blackwell on AI and education.
Recommendation No. 4 points to a children's rights impact assessment. I ask Ms Blackwell to expand on what she means by that and what it might look like in practical terms for the types of tools used in the education sector.
Ms Noeline Blackwell:
It is in the education sector and it is also wider. We are just trying to think of what could be done practically. What does not even need new legislation? One of the obligations on all public bodies at the moment is to respect and protect human rights and equality. That is their public sector duty. On foot of that alone and in line with obligations under the Convention on the Rights of the Child, any public body, including the education system, could identify where children's rights are engaged and whether they are protected either before or while deploying something new. They should put this kind of framework over it that just checks to ensure that children's rights are improved or at least are not adversely affected. There is not one single way of doing it.
The Deputy was talking about co-ordination with Ms Cronin earlier. I do not think it will happen but maybe we will manage to get whole-of-government approaches to everything; people are always hoping for that. Pending that, we could have a combined understanding so that, in the deployment and development of technology and its application by the State, including by Tusla and the education system, bodies look at its impact on children's rights. That is really what it is. I could bore the committee for hours talking about the methodologies, but I am sure the Cathaoirleach would not allow me. It is a way of identifying whether children's rights are engaged.
Darren O'Rourke (Meath East, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Do those tools and frameworks exist for those types of assessments?
Ms Noeline Blackwell:
They are something that the UN Convention on the Rights of the Child really promotes and there are several of them. The Children's Rights Alliance was responsible for one that was developed recently. It may not suit everyone, but the information is there and it is an existing obligation.
Lynn Ruane (Independent)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses for their presentations. We have been covering this for months. Questions were asked about Coimisiún na Meán and allowing the EU to take the lead in an investigation. On the other hand, we are waiting on the EU for CSAM. I understand the urgent need to have that audit of our laws to know what we can apply now instead of waiting. What is the expected date for the EU regulation? We will obviously need to bring in legislation to give effect to that. We have been talking about the end of 2026 or 2027, but it has already been delayed. Is that an accurate estimate? Will we actually see CSAM by those dates? What is the hold-up at EU level in the conversation around this? I hope that is not too wide a question.
Ms Noeline Blackwell:
I thank the Senator for that important question. The 2026 or 2027 timelines are really only about when it might have revised the EU AI Act, as we understand it. We cannot expect to see deepfakes and nudification apps prohibited until late 2026 or 2027 at the earliest. We have been given the date of late 2027 for the implementation of the high-risk stuff, but dates can be changed, as the Senator knows.
The CSAM directive and regulations have been meandering their way through the European Parliament, Commission and Council for a very long time. That is why our submission proposes that Ireland's Presidency of the EU be used as a way of driving it forward. We have also discussed this matter with ICCL and Ms Cronin may have something to say on it. There is an unfair and unreal polarisation between people who say that we have to preserve rights to privacy and people who say that children have to be protected.
Both are true. We believe there is no impetus to ensure that as that happens in the offline world - not perfectly, but it happens all the time - we can balance those things. It can also happen in the online world if we can get past the ideologues. That is my view.
Lynn Ruane (Independent)
Link to this: Individually | In context | Oireachtas source
Before I move to Ms Cronin, I want to understand more deeply the intersection of the ideology around data rights and privacy rights with the protection of children. Where is the crux of that tension, where one conflicts with the other, given that children's rights are also about data protection and privacy? What is the tension?
Ms Noeline Blackwell:
Our analysis is that at the European level, the rights of children to be protected really did not feature until very recently. The Parliament is looking at it right now. There is a good understanding in Europe about what data rights and privacy rights look like, and they have been through the GDPR. Honestly, if we look back through the debates on the topic, children's rights to protection from Article 3 harms under the European Convention on Human Rights - absolute rights - were not taken into consideration at all. Now, Europe does not seem able to find the mechanisms to say that children are entitled to their data rights and privacy rights, as well as their rights not to be subjected to very harmful abuse. In fact, what was happening right up to Good Friday or Easter was that even the very limited protection that the platforms were permitted to offer to try to identify CSAM ceased. It looks to us as if it was more ideological than a real effort to protect rights.
Ms Olga Cronin:
This is in respect of the child sexual abuse regulation, which has been logged at a European level for a number of years. The reason that has not crossed the line is that there has been a lot of debate about encryption specifically. Are we going to build pathways into encrypted messages to try to find child sexual abuse material? That brings a whole collateral intrusion issue. Are we going to set up scanning tools on all of our devices to try to scan material before it goes into that encrypted environment? That basically circumvents encryption but breaks the principle of encryption. Those are the issues that have kept that regulation from crossing the line.
Ms Blackwell and I have talked about this. Children's rights groups and privacy rights groups are being pitched against each other, whereas when we are able to sit down in a calm conversation, we can talk about looking for evidence-based solutions. Sometimes, a lot of the measures that are put forward sound like they are going to be a quick fix or a magic solution, but they are not. That has been the crux of that issue at a European level.
Lynn Ruane (Independent)
Link to this: Individually | In context | Oireachtas source
Have any evidence-based solutions been put forward?
Ms Olga Cronin:
It is pretty much an ongoing issue. I should say that there has been voluntary scanning by big tech for several years now. Ms Blackwell was referring to the interim derogation from the e-privacy directive, which has been allowing that scanning to continue. It should be said that the reason there has been an interim derogation is that there is no legal basis for that scanning, so there is a big question mark around the legality of that.
If it is of use to the committee, a couple of years ago, the ICCL published figures that we got from the Garda on this kind of scanning. It is problematic because of the number of false positives. If we were to mandate scanning across all communications, we would have already under-resourced gardaí and police officers across Europe-----
Lynn Ruane (Independent)
Link to this: Individually | In context | Oireachtas source
That goes back to the issue where the scanning revealed people's own children in the bath.
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses. I will start with Ms Blackwell. She may or may not have seen when Assistant Commissioner Willis was before the committee and gave evidence about live Garda investigations into the child sex abuse imagery generated through Grok - she was very open and clear that it was Grok - and, indeed, non-consensual adult intimate imagery. The evidence she gave at that hearing was that the Garda had commenced 244 investigations, that those investigations flowed from direct referrals from the National Centre for Missing and Exploited Children and that 50% of them related to child sex abuse imagery as defined in the Act. There were roughly 120 referrals of imagery that the Garda had identifiably proven was child sex abuse imagery. In a criminal investigation, the Garda is trying to identify the creator of that imagery but the assistant commissioner also acknowledged that, at least theoretically - she did not use the term "theoretically" - the platform could be construed under criminal law as aiding and abetting in that crime.
One of the things I am always somewhat concerned about in this discussion is this. While it is really important that the AI Act has been changed and will be changed to include these prohibitory practices, there will always be primacy for criminal law to deal with these crimes over any quasi-criminal sanction that could flow from Coimisiún na Meán or the Data Protection Commission. Was Ms Blackwell shocked at the number of child sex abuse images generated through an AI platform and under investigation? What is her assessment of the criminal justice regime in the area of protecting children? Are more things necessary at criminal justice level to address these issues?
Ms Noeline Blackwell:
I will take a bit of it and ask my colleague Ms Murphy to look at another bit of it. Missing Children Europe had a new report out yesterday. That organisation is the amalgam of all agencies looking at the welfare of missing children in Europe. It calculates that one in five children experienced online solicitation or grooming before turning 18. That is western Europe. Am I shocked? Yes. Am I surprised? No. Sorry, that is a formula but it is a formula that works.
AI and, in general, ICT have undoubtedly facilitated child sexual abuse in a way that increases the risks that were already there. It makes it much faster, more sophisticated and more international. Do we have the legislation we need? The audit that the Department of Justice, Home Affairs and Migration is undertaking right now is important in identifying the gaps in our criminal legislation. For instance, we would say that the generation of the AI and the tools to use it may not be enough. We would also support a recommendation of the online safety task force of the Department of Health that the legislation we are putting in place right now should allow for directors and senior executives of the platforms to be held criminally liable where harm happens. It happens under the Data Protection Act but it does not happen under the tech legislation.
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
Would Ms Blackwell draw a distinction, even at the level of principle, between a platform on which grooming has taken place and an AI platform that generates an image and whose proprietary technology is capable of generating such an image? For example, in a grooming incident, there might be a user on one side and a victim on the other and there is a medium being used. Is there a distinction between those two things?
Should there be a distinction? What kind of criminal justice regime should we have? Should it differ for both those things?
Ms Noeline Blackwell:
My analysis is that we would look at the harm. Whether you use an AI-generated tool or other methodology on a platform, if it results in criminal harm, then that is where we should look. I do not think a distinction should be made between them. Some of the social media platforms aimed at very young children - not just social media, but gaming platforms and social media platforms - are engaged in trying to groom children. It happens across non-AI-specific platforms but AI is increasing the risk of children being groomed online.
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
I was somewhat surprised when Assistant Commissioner Willis said that, at least at the level of principle, an AI platform cloud be viewed as aiding and abetting a crime in this instance. Like has been said, there are horrific cases of online grooming taking place on a multitude of platforms. While I am not aware of any, are the witnesses aware of any of those platforms ever being criminally investigated under Irish law for permitting their systems to be used for grooming or just as a result of their systems being used for such? Are there such examples?
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
I do not want to name a platform, but let us say there was a messaging platform being used by a criminal on one side to message a victim. I think that what Assistant Commissioner Willis was getting at was that Grok, theoretically at least, could be construed as aiding and abetting the crime of the user generating child sex abuse imagery, whereas in a grooming situation where there is a criminal using a messaging service to message the victim, I am not aware of the messaging service being investigated for permitting its use for that harm. Is there a distinction between those two things? Should a distinction be made? Do we need to strengthen our criminal laws in respect of both of them?
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I ask the witnesses to be as succinct as possible.
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
Yes.
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
Was that on X?
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
I think what Assistant Commissioner Willis was suggesting was that the AI platform itself was the thing that could create the image whereas the person typing in the messages is the person doing the thing, if the witnesses know what I mean. There is probably a distinction there that needs examination.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
We may come back on a second round.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
I thank everybody for being here. I want to dive into some of the points that Ms Cronin made in her opening statement and get into the details of what did and did not happen. As I see it, a nudification app was launched on a more mainstream platform than had previously been the case. There was an outcry and scandal and we still do not have any actual accountability. Nobody to my knowledge and, I believe, Ms Cronin's knowledge has been held to account for this, arrested or prosecuted. I appreciate that there are multiple investigations and initiatives under way to try to deal with it but that still is the reality. If Ms Cronin knows anything to the contrary, please do not let me know.
I want to get into what is effective within our existing legislation. I agree with Ms Cronin in wondering why we would need to create new laws when the enforcement of existing laws is not happening to the right degree. Ms Cronin said that she thought Coimisiún na Meán should have launched an investigation under the online safety code but that it did not do so or that it believed the threshold had not been met to warrant that. Will Ms Cronin tell us more of her analysis of that? Was it the right call? Does she think Coimisiún na Meán still could launch an investigation under, for example, the online safety code?
Ms Olga Cronin:
I am sure it probably still could act. Obviously, this is potentially an ongoing issue.
In relation to the OSMR Act, I can share this table with the witnesses again, if they would like. The Act does not define what is illegal, but it provides this framework for Coimisiún na Meán with regard to illegal content online via the OSMR and the online safety codes, specifically video-sharing platform services, including X. It is our position that it could have acted with its powers. Under the OSMR Act, Part 8B of the Broadcasting Act 2009 provides that where there is a reason to believe that there has been a failure to comply with an online code - otherwise a contravention - an authorised officer appointed by Coimisiún na Meán can carry out investigations. This is under sections 139ZH and 139ZI.
The online safety code states that video-sharing platforms must take appropriate steps to protect "the general public from ... user-generated videos ... containing content the dissemination of which constitutes an activity which is a criminal offence under Union law", including child pornography. That is section 10.1.c of the code. They must also take appropriate measures to protect children from programmes or user-generated videos which may impair their physical, mental or moral development in accordance with Article 6a(1) of the audiovisual media services directive.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
I do not want to go any further into it because it will eat up the time but I really appreciate that because it seems to me that my second question is redundant. The online safety code does not need to be addressed or amended because it seems to offer the powers and the initiative to Coimisiún na Meán to investigate. An coimisiún just chose not to. That was essentially an organisation-level decision.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Of course.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
It also decided or deemed that EU-level co-operation was more appropriate, but I do not understand that on the basis that there is no reason we cannot do both. There is no reason it could not co-operate with Europe while also looking at a domestic-level investigation.
I have a similar question around DPC because it eventually launched its investigation under section 110 of its relevant legislation, whereas the ICCL's recommendation had been under section 134. Is that right?
Ms Olga Cronin:
Yes. We said that it could have taken that action under section 110, which is to open an own-volition inquiry. There are circumstances under Article 34 that it can make an application to the High Court for the suspension and restriction of processing of data, and that is where there is an urgent need to protect data subject rights and freedoms under the 2018 Act.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
That was a choice for it.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Did it also ultimately look towards the more European-level solution or was it at domestic level?
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
It thank Ms Cronin.
I will pick up on something that Deputy Geoghegan talked about. In terms of the separation between AI and platforms, as far as I am concerned, the more we progress with this - we are dealing with this matter at this committee but also at the committee on culture, communications and sport where discussions on the regulation of online platforms are held - the less distinction there is between social media platforms and AI tools. For example, the pig butchering and other scamming that is happening on a massive scale right now. In these scam farms, units deploy filtering tools to identify victims. They then groom them, close the deal and get into the crypto sequence where they basically take all of their money. They are not afraid of using AI tools. I realise that it is a scam industry, so it is an illegal industry in and off itself, but I have worked in the tech sector and the idea that a tech company is not going to use an automative tool, which "artificial intelligence" as a term essentially covers, to improve the efficiency of its products is just ridiculous. That is what they all do. Is that witnesses' understanding of this?
Ms Alex Murphy:
From the outset, artificial intelligence has been embedded in some form across all social media platforms. That is what drives our algorithms. It is the basis behind them. What we are seeing now is this huge escalation of generative AI and that is why we are seeing a massive increase in the dissemination of child sexual abuse materials. All of these harms have been lifted to another level because AI has escalated to another level. For us, what we are predominantly seeing is that AI systems are embedded across social media and gaming platforms - anything digitised.
Education was traditionally digitised but now here we go, we are moving into a digital age and artificial intelligence is in the very epicentre of that. However, it is not the be-all and end-all. It is what is exacerbating harms but these harms were already there.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
If I hear the term "human in the loop" one more time I will go ballistic. That seems to be the catch-all - "Yes, but we have a human in the loop". I am sorry, but that is just not enough.
I have one final question. To me, Grok is one feature on one app on one platform that has created untold harm in society. Are there features or tools on the horizon that are operating on the dark web, which was where notification was happening before it emerged into the Grok space and the more mainstream space? Are there any other tools or features that are essentially unnerving or worrying the witnesses at the moment?
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Is that AI girlfriends?
Ms Noeline Blackwell:
Yes. AI girlfriends and, on the nudification things, as one of our colleague organisations CyberSafeKids highlighted, what about people in bikinis, which are not intimate images? Certainly, AI girlfriends is an issue. More and more, as we mentioned in the submission, chatbots are leading children into terrible harm, including suicide, as happened recently in the UK.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Yes. I have raised that a number of times. I thank the witnesses and apologise for going over time.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses for coming in this morning. I will start with Ms Blackwell and Ms Murphy. In the statement they said a child rights impact assessment has to be applied in development and deployment of AI to ensure there is no danger to children. Who should be the assessors? If it is not complied with, how is non-compliance dealt with?
Ms Noeline Blackwell:
At the moment, not having anything there to deal with such non-compliance is an absolute scandal. The reason we are suggesting something like a child rights impact assessment is it is something that can be built into the smallest policy, the biggest policy or a piece of legislation. It can be built into an EU Act, if they wanted to. If it is included, it gives the person who is going to deploy the tool or the policy, or run the class or be the child protection social worker, an understanding of what the risks are to children's rights. Once that is recognised, it would be expected in a country that has ratified all the rights conventions and says it obeys the rule of law, that it would engage with children's rights and ensure they are protected. If they are not, then the law will very often say, "You knew. Not only did you violate children's rights but you knew you were doing it."
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Who does Ms Blackwell actually see as the assessors?
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
How do we make sure they are accountable?
Ms Noeline Blackwell:
One thing that is missing from this area, and this goes to a previous question that was raised, is about how we know victims are looked after. There is no independent complaint system whatsoever. How we would look after it is we would use the law. We would have independent complaint systems. There would be ways to ensure compliance.
Ms Alex Murphy:
Under the Digital Services Act, very large online platforms have to perform risk assessments anyway. We argue that a child rights impact assessment should be built into this risk assessment process. Under the EU AI Act as well, risk assessments are required for AI providers. Again, we would say that child rights impact assessments could be built in based on this model. It would be the providers of the AI system or the providers of the online platform that are initially doing them.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Is AI trustworthy and capable of discerning age? How do the witnesses feel about AI using biometrics on children? For example, some schools are using it for attendance tracking, for meals and for library books, etc. Does this infringe on children's rights?
Ms Noeline Blackwell:
That is one of the things that relates to the tech areas in schools. There is an awful lot of information being collected about children through school programmes that are often seen as very beneficial for digital literacy or whatever. Children have the same rights as anybody else to the protection of their data and privacy. Is that being properly observed? We do not see how it could be in the absence of good licensing systems.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
We know of the European Commission omnibus regulation on AI, but it is proposing to weaken GDPR, particularly for children. Does Ms Cronin believe this should be prevented or even pushed back?
Ms Olga Cronin:
I might go back to something the Deputy said about the biometrics of kids in schools. If there are schools in Ireland that are using facial recognition or biometric scanning for attendance, the library or food, that would be deeply problematic because I do not see how it would be necessary and proportionate. The Data Protection Commission would also be very interested to hear if that kind of thing was happening.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Sorry. It is definitely being used in other countries. That is what I am referencing.
Ms Olga Cronin:
I understand. I thank the Deputy. On the simplification process, the ICCL has already raised its concerns through my colleagues Dr. Johnny Ryan and Dr. Kris Shrishak. Weakening data protection laws is not a good idea. They are there for a reason and we should be very protective of them.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
We spoke about the Grok case but what is Ms Cronin's opinion on the Data Protection Commission not making one single inquiry into Google despite repeated complaints?
Ms Olga Cronin:
This is of concern to the ICCL. My colleague, Dr. Johnny Ryan, would have talked about this extensively. The enforcement of our data protection law is extremely important, not just for people here in Ireland, but for everyone across Europe, adults and children alike. It is very surprising that the DPC has not concluded any investigation into Google.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I have one more quick question. Workers in the industry have to look at a lot of videos that sometimes involve the abuse of children and horrific stuff happening all over. Is there any way those workers could be protected or could there be built-in protection in the industry?
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Does Ms Cronin think they are there at the moment?
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I thank Ms Cronin.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I thank Deputy Mythen. It now falls to me. I will thank our witnesses again. Given the discussion, it might be useful if the committee wrote to Coimisiún na Meán to seek an update from it on the investigations we have heard about here. I appreciate that Garda investigations may be going on, but it might be useful if we wrote to the Garda to get a more general update because that would reflect some of the discussions.
I welcome how Ms Blackwell talked about individual directors' liabilities, but during the discussion on the OSMR Act by those of us who sat on the media committee at the time, the idea was not one that gained much currency, unfortunately. I hope we will be able to revisit it. I do not believe fines work. It is important that we hold individuals responsible for actions.
I want to turn to the whole area of the ban on child sexual abuse material, nudification apps and so on. I pay tribute to Michael McNamara, MEP, who has done an awful lot of work at European Parliament level on this and on the discussion around the AI Act and the omnibus. My understanding is that, as and from December, they will be banned on all platforms. I take it that is the witnesses' understanding as well. We will be in a situation. Are there any concerns that the ICCL has around what is coming into place from December?
Ms Olga Cronin:
At this moment, not particularly. There was something like maybe 55 nudification apps on the Google Play Store and 47 on the App Store. We do not have a problem with the banning of nudification apps. The issue is that most general-purpose AI models can create this kind of material. That is the real problem with nudification.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I accept that, but I think it is moves at EU level in terms of non-consensual use of individual images-----
Ms Noeline Blackwell:
The best-case scenario is December next, but it could run into 2027, which is why recommendations from committees such as this are so helpful in keeping a bit of pressure on, especially with the Presidency coming up. The more problematic stuff is the deepfakes and other AI applications like that. You cannot just say you will ban them all or prohibit them all because some of them will have capacity to do things but will have other capacities as well. They are not even seen as high risk as it stands, so bringing them into the high-risk category would be very important, but so too then would be the capacity to keep reviewing what is high risk. When the EU AI Act said that they were limited risk, which it did at the time when it was passed, the technology was not as good as it is now. All these things need to be kept under review, and somebody has to start asking children and people who advocate for them what are they seeing and the things that need to be taken and brought into high-risk categories. A lot more needs to be classified as high risk. Even the high-risk categories will not be there until 2027, maybe the end of it, and not all of them because if they are associated with, say, the health sector, they will go into that sector. There is just something very weak about the way in which people are seeing very damaging material and getting it moved out of the system. The current clunky system is not able to manage.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
My view is that we need to look at an individual's digital identity, including their voice, their likeness and their image, and that it needs to be protected, not just for children but indeed for members across society more widely. We start to assert our own digital identity rights in order that my digital identity cannot be used by a company-----
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
-----without my permission. The Government will be making online safety a priority during the EU Presidency. In terms of recommendations from this committee and what we talked about with regard to the AI Act, are there any other elements you think we should focus on during the Presidency? The witnesses will be aware, obviously, that there will be an AI summit in October, which this committee will have an input into, and that themes are being discussed at present. What themes would the witnesses like to see explored?
Ms Noeline Blackwell:
When the general scheme of Ireland's AI Act was being discussed and we were giving witness with the Irish Human Rights and Equality Commission, Mr. Liam Herrick mentioned that at that summit, for instance, it would be really useful to have a focus on human rights in that area, not necessarily a shadow but just a focus within it on those issues, and perhaps invite him, but we would say invite us as well. Some of the recommendations are at that policy level that could be done immediately, building up that understanding of children's rights and impact assessments. That is something that is known throughout Europe but during the Presidency, where Ireland will have the heft and the persuasive power will be in bringing forward the CSAM regulation and the directive. It is just extraordinary that those two things are not there. The CSAM directive relates to those who are 12 or 14 years old. It is completely out of date. There is no platform accountability at all at the moment. There is such capacity to build the better evidence base that Ms Cronin and I were talking about and that we will continue to talk about at the NGO level, but if the Government were to do that, it would be a massive-----
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
That is fair. With the committee's agreement, we might raise that further and in a direct manner.
I am conscious of the time. If others want to come in on this, they should feel free to do so. It is important that Ms Blackwell mentioned informed consent of children.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
In the era of AI, can Ms Blackwell define for us what she means by "informed consent of children"?
Ms Noeline Blackwell:
I will have to do another paper on that. In many ways, the concept of consent is becoming much better known and understood. When we talk about consent, we have to keep putting in "informed", which means understanding the risks. On AI, we keep talking about the platforms as if that is all the children are on. They are on side-----
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I am thinking about Roblox.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
For a child at seven or eight-----
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
-----as opposed to a 15-year-old, it comes back to discussions we would have had.
Ms Noeline Blackwell:
It again goes to the other committee of which the Cathaoirleach was a member, the media and communications committee, which identified, for instance, the classification system, which our colleagues in CyberSafeKids are very strong about, and which the online safety task force also recommended. Those are the kinds of things that are from the top level down. At the level of children, they have a right to know what they are consenting to. That is where there is a lot of work to be done. You need child-friendly language and an assessment of what the children are consenting to that is plainly done. Information has to be available to their guardians and caregivers and there has to be a digital duty of care towards the children.
Laura Harmon (Labour)
Link to this: Individually | In context | Oireachtas source
I apologise for missing the beginning of the meeting. I was at another meeting. I read the witnesses' statements.
We saw the research that was done and published by UNICEF, which identified that there were 1.2 million children across 11 countries whose images were used in sexual deepfakes in the past year. Two thirds of children are worried that their images could be used in that way. Do we in Ireland need to do more to collect data around what is happening and who should be responsible for that?
Ms Alex Murphy:
One issue that is arising is how we are collecting the data and where it is coming to. Hotline is one of our biggest reporters, but there is a difference between reports they receive that are referrals and action reports, which turn out to be CSAM. There is also the fact that the online world is not territorial, per se. Many of the reports can come from different countries or within Ireland and it is about trying to assess that on a country basis and on a western European basis. Ms Blackwell mentioned earlier what came out today from Missing Children Europe, which was that one in five children in western Europe alone have experienced grooming and solicitation online before they turned 18. As more international hotlines, such as those at the National Center for Missing and Exploited Children, NCMEC, and the Association of Internet Hotline Providers, INHOPE, collaborate, it is easier for data to be collected and put together. However, there is still a massive gap. So much is happening on the dark web or in the unregulated area of the digital environment. It is a lot harder to get data from there. While we are seeing huge numbers, we are still missing a large portion. Especially now that the CSAM e-privacy derogation at EU level is no longer facilitating platforms to voluntarily detect, we are missing a whole other portion. The data we are seeing-----
Ms Noeline Blackwell:
On that particular issue, it is the lack of transparency of the companies. We do not know how many complaints are going into the tech companies or how they are being dealt with. We do not have an individual complaints mechanism even though legislation provides for one and Coimisiún na Meán says it is working on some version of one.
The Dublin Rape Crisis Centre is launching its annual report today. It has seen an increase in the number of people who are reporting rape and other sexual abuse. That is, the centre will say, only a tiny percentage of those who actually experience it. You need to collect your data in several ways. The companies know more than they are telling us. There is no individual complaints mechanism. You can complain to Coimisiún na Meán, but all it will do is to use that for its investigations. That is no indicator of what is there. There are no simple ways for children and young people to report safely, except through Hotline, which can take something down.
Therefore, an awful lot of work needs to be done to provide safe and effective mechanisms for people to identify and get help and support if they are the victims of any sort of violations, and then to be able to report them and access justice.
Laura Harmon (Labour)
Link to this: Individually | In context | Oireachtas source
What strikes me is that there is a real need for legislation and regulation. Also, a holistic education approach is needed in relation to policing, but also in relation to young people themselves, their parents and guardians, teachers and our education system.
We had a really interesting session here a number of weeks ago with experts from Estonia on their use of an AI programme in schools. I wonder whether more needs to be done on education. Does Ms Blackwell want to elaborate on that?
Ms Noeline Blackwell:
Finland is always showing up as doing well in this area. Children's understanding of the digital world is so much better as a result of a graded approach to it over their entire lifetime. Digital awareness and digital literacy are key for children, their caregivers and their parents, provided it is understood that parents and caregivers are not a homogenous bunch and that many of them barely manage to survive. It cannot be left to them to build awareness.
A really good example people often mention is the threat to share intimate images. There was a really extensive campaign on this across social media and all sorts of other media, to the extent that almost everybody in Ireland understands it is a crime to threaten to share an intimate image of somebody, as the Garda will confirm. A campaign can be successful. There have been many campaigns that did not work. It goes to show that children cannot know if their parents do not know. There is no focus on this. Teachers will say not to suggest teachers have to do more, because they are at the pin of their collar with curriculums and so on, but this is not a matter of a subject. Everything in a child's life is now being handled digitally and there needs to be quite an uplift in the way this is addressed. It is like trying to teach children to write with a scratchy pen or a quill. There is no point in trying to do that anymore. The same applies to digital awareness. In this regard, the Senator is certainly right.
Ms Alex Murphy:
To add a small point, a holistic approach is definitely necessary, but we do not want to see a digital divide created among children whereby children from certain backgrounds are left behind in this digital move that is already under way. We want to make sure all children benefit, as they have a right to do under the UN convention. We want to make sure this holistic approach is for everyone.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses. We are going to have a second round. Again, we have seven minutes each for questions and answers.
James Geoghegan (Dublin Bay South, Fine Gael)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses. Not for the first time in her contribution to public life, Ms Blackwell nailed it with her point about awareness and the difference an awareness campaign can make. The campaign she spoke about was exactly what I was going to raise with her.
One of my concerns in all this discussion is that while we keep talking about Coimisiún na Meán and the Data Protection Commission, which are very important entities, we are ultimately talking about perpetrators committing crimes and about people knowing these offences are crimes. One of the more shocking elements of the Grok issue was that, for a short period, it created a permissive environment for behaviour. As I understand it, the Grok app became the most downloaded AI platform. It went right to the top of the list of Apple App Store in Ireland and other jurisdictions.
The mention of a campaign gave me hope. There should be campaigns about what are and are not crimes, in particular nudification. It seems insane that we would have to have awareness campaigns about child sexual abuse imagery. I fear there is a glacial response on a regulatory criminal level, whatever way one wants to look at it, compared with the incredible speed of AI development. We see that in many different areas. Ms Murphy put it very clearly. At the end of the day, a lot of the stuff that is taking place was already happening on these platforms and is boosted by AI. A well-run campaign has a function.
What more can we do to protect our children and adults from the turbocharging of harms that are already taking place in the online space? The witnesses may have heard that the child actress from "Matilda" has become somewhat of a spokesperson on this issue. Later in life she discovered pornography of her as a child actress and it put her off acting. She did not speak about it for years. Her big fear is what happened to her as a famous child actress being turbocharged everywhere.
Investigations are happening and the AI Act is moving. It was positive that it was not marred by a red tape free speech debate. There was an actual response to a civic need. How do we keep that momentum going? How do we keep it on the agenda? How do we ensure our criminal justice system has sufficient tools as well as all of the regulatory changes? What do we need to do? I do not expect the witnesses to have all the answers. What more do we need to do to keep the civic response, regulatory and criminal justice responses turbocharged?
Ms Olga Cronin:
I commend the committee on having representatives of all of those groups before it and asking them questions as to how they have acted. In terms of keeping this area under the spotlight, we just have to do it. All bodies have to do it and we have to hold them to account. This is not easy. There are no quick solutions or magic bullets. It is a continuous battle. We need to use the tools and powers we have under different legislation, notice when there are gaps and respond in a rights respecting way that does not create collateral intrusion or set up systems of surveillance or censorship. We need to use the tools we have, spot the gaps when they exist and act accordingly.
Ms Noeline Blackwell:
We keep complaining about how we are in Europe. As regulation became less important in other parts of the tech world, in particular the United States, it was said that people felt they had permission to do things. I think the tech companies felt they had permission to do more when America said it would have less regulation. We need to recognise the value of Europe staying in some sort of a rules-based system. The simplification of legislation, omnibus legislation and trying to get rid of duplication, which is a good thing, runs the risk that we will forget about rights and obligations. In the case of AI, Ireland has to balance the thrust for innovation and leadership in one area with the fact that, in order for the rule of law to apply, the tech products we use have to be safe.
As Ms Cronin said, that is not one piece of the work. I will add to the frightening comment on child sexual abuse material. A lot of the nudification apps are particularly difficult to prosecute because kids are using them to abuse, bully and laugh at other children.
We are looking at how we actually build not just the awareness I was talking about when I was answering Senator Harmon but how we build their understanding of what decency and respect is, how they speak to each other and how they have to behave in schools. We are using a whole different set of tools now than were used 20 years ago. They are moving very fast. Every time we write a paper or do something, it is out of date by the time it is published. Members here probably find the same thing themselves, Chair.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
We do, yes.
Ms Noeline Blackwell:
Nonetheless, we are coming around to the same thing. We still live in a world where people are entitled to use safe products and for their rights not to be violated. In some ways, we have to build up more systems than we have right now for that. The committee is inviting us in; that is terrific. There is a need for regular recognition that the more vulnerable people are in this world, the more these new tools will put them at risk. That has to be managed as well.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
Ms Cronin wanted to make a brief point.
Ms Olga Cronin:
It struck me that in respect of this question about nudification tools and kids using it in respect of other kids either for maybe fun - they think it is funny or whatever - but also for more intrusive and dangerous reasons, there is some research that has been carried out in the US that might be of interest to the committee by Ms Riana Pfefferkorn from Stanford University. They are kind of ahead of us on this, let us say. That might be of use to the committee when members are creating recommendations in respect of that child issue.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
I want to pick up where we left off on chatbots. I have raised this issue a number of times, and the suicide risk they pose. I do not think it is particularly young people anymore because, from what I am reading, it seems to be across all age ranges. I have quoted the tragic case of Adam Raine, which is currently the subject of a court case that his parents have taken against OpenAI and which is ongoing. The case essentially states that Adam started using ChatGPT in late 2024. He was 16 so, potentially, would not have been affected by any age verification ban. He asked for assistance with homework. Six months later, he died by suicide. It was only in the weeks after his death that his parents discovered what were chat logs, essentially. Part of the defence was around how the sycophancy dial had been turned up in this case and that the company was learning from that and would address it in future. However, what that says to me is that, at the moment, social media platforms, for example, want our eyes on the screen and they use addictive technology to do that, to essentially keep our eyes on the screen. What chatbots and AI tools more generally, or the AI girlfriends that were mentioned earlier, are seeking to get from us is not just our attention but our emotion. It is a more sinister and deeper connection that they are creating with users to hold us online to, again, replicate that addictive behaviour, but on a deeper level. The scams I mentioned earlier speak to things like loneliness and lack of reputation, so when they are looking for people, they look for people who are looking for business opportunities on some platforms, for example, or looking for love or friendship on other platforms. They then identify that those people are feeling vulnerable and eat into those very human, very vulnerable emotions to create a connection, which they then exploit for gain. That one is criminal; we know that. The other one - the chatbot use - is not criminal as it stands but is, obviously, an area that is under investigation. Honestly, I do not even know what my question is here. If I go to Ms Cronin first, perhaps, within the existing legal framework, are chatbots something she thinks we can actually look at dealing with at the moment or is that going to require new legislation?
Ms Olga Cronin:
I am not entirely certain but with regard to the AI Act, when it comes to general AI systems, perhaps there can be a sycophancy risk assessment under the risk assessments that have to be carried out under the systemic risk section. I am not entirely sure. There has been some research about this very thing by a number of academics in the UK. The name of the report is Invisible No More: How AI Chatbots Are Reshaping Violence Against Women and Girls.
That might also be of interest to the committee. The Deputy is completely correct. That sycophancy issue is deeply troubling. It works in the same way as the algorithms. I agree with the idea of not trying to attract the eyes but the emotions.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Do the witnesses have anything to add?
Ms Alex Murphy:
I thought that was subtle. We have said it in a few pieces as well, but we would really argue that chatbots should be deemed high risk and that certain exceptions could be deemed limited risk. Ultimately, we believe they should be considered high risk. It is not only suicide that they are pushing forward. It can be inappropriate and sexualised conversation or eating disorder content. It is exactly the same as the algorithms, except that it is an AI system having a conversation with a person and consistently convincing them of something.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
That creates that deeper connection.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
In that case, the specific things were the chatbot offering to write a suicide note, essentially saying to this young man, "I see you in a way that nobody else sees you" and "I see you at your most vulnerable", and almost praising and admiring his exposure and vulnerability to it. That is exactly what it is trying to do - develop a connection that creates a dependency such that a person is only going to open up to it.
Ms Olga Cronin:
There has been some discussion, I am not sure where it stands at the moment, about the idea of certain chatbots being treated like very large online search engines. If that were the case, it would bring it under the remit of the DSA, which would mean that the relevant risk assessments and risk mitigation measures would have to kick in. However, I am not entirely certain as to where that conversation stands. There have been some discussions about the matter.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
While I understand the legal segregation of very large online platforms, VLOPs, which is allowing us to build this framework, does that not create a black market tendency or a tendency among smaller platforms to act in a particular way. When we look at what is happening with Grok, we can see that once it is harder for people to access, all of these smaller, criminal kind of players that are less concerned about regulation are going to be putting them out. We are going to see the same in the chatbot space. Is that a risk?
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Do the witnesses support the separation of the VLOPs?
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
There was a lot of information put out at the time of the Grok scandal for people to go to the police with individual cases. Are people still coming to the ICCL?
Ms Noeline Blackwell:
There are more than 165 members in the Children's Rights Alliance. The harms to children digitally and the real harms from bullying to sexual abuse really bother parents groups, children's groups, teachers and people who work with vulnerable children or children of different ethnicities.
Laura Harmon (Labour)
Link to this: Individually | In context | Oireachtas source
My question relates to the witnesses' views on AI smart glasses. There were reports on this recently across the country, particularly in Cork, where a woman was recorded in her workplace and another was recorded receiving medical attention. Is this on the witnesses' radar? Is it something they have thought about? My understanding is that Snapchat is bringing out its own version of Meta glasses. Potentially, this would have a younger audience. I see a lot of risks with that in terms of people not being aware that they are being recorded.
Currently, it can be difficult to tell if you are being recorded even with a phone. This is another level, however. These smart glasses have a light that comes on to show when they are recording, but that can be covered. In terms of privacy, is there more legislation or regulation needed in this regard?
I want to make a broader statement. We obviously know that sexual violence continues to be an issue in society. Paedophilia is also an issue. Overall, is it fair to say, from the point of view of the ICCL, that this has been compounded by AI and technological advances? Is that making it easier for predators and perpetrators to operate and to commit violations? Is it compounding the issue?
Ms Olga Cronin:
I will speak to the point about smart glasses. They are really creepy. There is nothing else to say. In other jurisdictions, we see them being coupled with facial recognition, which is deeply problematic. I am not sure it can get any more intrusive than that. With the smart glasses, there was a different company that tried to bring them out many years ago. I think Google tried to bring out smart glasses, but the project just went by the by. The Data Protection Commission brought out some guidelines in that regard. There is an idea that the light being on to notify people that they are in use is supposed to be a safeguard. As the Senator said, however, you may not see that light. The glasses are deeply problematic and intrusive. I wish we could make everyone feel bad about getting them in order to stop them, because they have no place in Irish society.
I remember being contacted by someone who wanted to use a particular pair of glasses that could see for someone who had an eyesight problem. Those glasses were able to tell the person what they were seeing. That is totally different from the smart glasses the Senator referred to, which are problematic.
Ms Noeline Blackwell:
If the Senator is asking, in the context of AI, whether the digital age is making the issues of abuse, serious harm and violation of rights worse, the answer is that it is because it is making it faster and easier. It is harder to see where the societal limits and taboos are than it was previously. We have to adjust to it, and we are just not doing so quickly enough.
Laura Harmon (Labour)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I have two brief questions for our three witnesses. What is their opinion on turning off recommender algorithms? In the context of the main question in the public domain at the moment, what are the witnesses thoughts on banning under-16s from social media?
Ms Olga Cronin:
The ICCL has been extremely vocal on the recommender algorithm issue for years. We think they should be turned off by default. We think that under data protection law they should not be turned on, particularly in the context of the manner in which they use our personal data. Our position has been clear on that for many years.
On the question of a social media ban for under-16s, I do not think measures like that are effective. Sometimes, the goal that is being put forward is not very clear. Age verification systems are circumventable, but also can be really intrusive in how they are supposed to be implemented. The ICCL and Digital Rights Ireland have said a lot about the plans for age verification or social media bans for under-16s in Ireland. We also think that the plan is problematic, or, at least, what we know of it, in that it is connected to the public services card, which brings a whole host of other problems because there are legal issues with that card in and of itself.
Ms Noeline Blackwell:
We absolutely agree with Ms Cronin and the ICCL on recommender algorithms for children. They should be turned off by default. Another Oireachtas committee has made that recommendation last week.
On the social media ban, we take the view, as Ms Cronin said, that it is not effective. Neither do we think it is the right way to go about things.
We absolutely accept that there is content that children should not be seeing. Children should be protected from that. Social media companies and all other platforms should be making sure and doing their level best - we believe they are not - to make sure that children. cannot access content that is harmful for them. They just should not be able to do it. Everyone has a different idea. Australia is the only country that has brought in a social media ban. Loads of countries across Europe and South America are proposing it, but they all have different ideas about what it would look like. The thing about it is that it stops children from having a social media account. It does not stop them from accessing the Internet. They can still look at their YouTube or whatever it is on their browser. They can, of course, as seven out of every ten children in Australia have done, get around it very quickly. That is the reality of the way it is working. There is no evidence base that it works and there is no evidence base that children are safer because children are either going to get around it or they are going to move to channels that have even less protection on them. If the social media companies are saying there is nobody under a certain age on their platform, they have even less compulsion to do anything that is right.
The other thing is that it is highly discriminatory. The Deputy comes from a constituency with a lot of rural people in it. If you live half a mile outside a town, your child is getting on a bus from school and going home and that is the end of their activity. You cannot say to that child to go out to the playground and play with the children in the courtyard, because they are not there, so it is discriminatory. Children who, for instance, might want to identify with other children who are, let us say, gay cannot do so if they do not have the social media accounts. They do not have a way of asking the questions about it. Children who have a disability perhaps cannot go out either. They feel included right now, so it is discriminatory and therefore wrong from a children's rights point of view. We would also say it is ineffective.
We do not fully agree with the ICCL that we cannot use age verification. We think it is a tool. It is not the best tool in the box but there has to be a privacy-preserving way of the Internet better understanding the age of children. I am not saying the tech companies could not do it. I am saying they do not seem to have any incentive to find it themselves, so it does look like states and others are going to have to come up with a system. Probably the best thing coming up is the EU Commission's and the Irish shadow system which we have to hope will give a better understanding of when a platform is encountering somebody under a certain age.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I thank Ms Blackwell. My last question is for Ms Cronin. I know this is off the subject. This morning we had the bank, Standard Chartered, just announcing that it is letting off more than 15% of its staff. Meta said it is going to let go at least a couple of thousand in the next month. Amazon has done it and Oracle has done it. How can workers be protected, especially those workers who are working in the public service? It seems to be the trend from the big tech companies. They are letting off thousands of workers overnight.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Yes, thousands of jobs.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
Is Ms Cronin worried about the situation?
Ms Olga Cronin:
When it comes to things like content moderation teams and trust and safety teams being hollowed out, which we have seen, that does have a direct implication for people, for users of the systems and platforms, because there is a knock-on effect on their experience with the service. I would obviously have huge sympathy for people who lose their jobs.
Johnny Mythen (Wexford, Sinn Fein)
Link to this: Individually | In context | Oireachtas source
I thank Ms Cronin.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
We will have sessions around AI and employment, and we are going to move on to that. I found some of the discussion quite interesting, particularly in the context of chatbots coming under the DSA. We can certainly explore that. Part of this comes back to the fact that we should always look at the technology. It is not a goal in itself; it is an enabler. There are lots of examples of where - for children and young people, Ms Blackwell gave the example of social media - AI will be able to help children to learn and to integrate. There are a lot of positives in that regard. Similarly, Amazon has partnered with Vision Ireland in respect of AI-enabled glasses for people who are visually impaired. As with any technology, it is down to how we use it. We are always going to be trying to catch up when it comes to regulation. However, it comes back to society's views.
Deputy Geoghegan made the point that it is surprising to have to tell people not to engage in respect of child sexual abuse material. Prior to AI, it was there. AI is just a new tool. As Ms Blackwell said, it is fast and we move on. Things are moving so quickly.
I am a big fan of public information campaigns such as campaigns relating to intimate images and consent. I am old enough to remember the advertisement with the line "I don't know what a tracker mortgage is". I still think that campaign was hugely successful in terms of people understanding financial literacy. We should run campaigns with the lines "I don't know what an algorithm is", "I don't know what a chatbot is" or "I don't know what a deep fake is". These would be aimed not only at children and young people but also at parents. Part of the challenge is about how we talk in society with children and young people. Many parents do not necessarily feel confident enough to engage with young people about their online behaviour or how they use AI. Equally, it is a challenge for children and young people. It is about how we, as a society, equip people with the necessary tools. I am always struck by the fact that school boards of management that have to draw up bullying policies in this regard. In the past, it was much easier when there was bullying in the school playground, whereas now most of the bullying happens online. As was mentioned, a lot of the bullying now is in the form of AI-generated material, which causes a problem. If the witnesses were to make recommendations around what we need to do on the education side and on the public information campaign side, what would they like to see happen? Who should be doing it? I appreciate that they do not want to say that teachers should be doing it.
Ms Olga Cronin:
Publicity campaigns around people's rights and how they can go about vindicating them are useful. When it comes to illegal content or the OSMR Act 2022 and the DSA, it is not always clear to people what they can and cannot do. There might even be expectations of bodies that actually do not have the particular means or power to do certain things. Media literacy was one campaign. Another that has probably been quite successful is the Data Protection Commission's Pause Before You Post campaign, which deals with images. That feeds into this, because we are talking about images that are used by others. More "know your rights" guides would be useful.
Ms Noeline Blackwell:
My favourite of all time, I wish they would resurrect it, is the one that teaches you how to go around roundabouts when driving. They changed the rules around that. That bothers me. It used to be great when they used to run these little things about roundabouts. To be fair to Coimisiún na Meán, it has produced some really good material. It is the regulator, and this is going to be a bigger piece of work. It has always taken a champion in government to run these campaigns and to understand the benefits of them. In some ways, these will have to be championed. They will also have to be constant because there is the "know your rights" aspect, there is the "What is an algorithm?" There is also "Who is Claude?", as somebody said to me the other day.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
Who is the champion? Is it Coimisiún na Meán?
Ms Noeline Blackwell:
It might very well be, but it would need different resourcing. It has a communications team, but that is to communicate in regulator's language. Coimisiún na Meán is the right place to put it because someone whose job is to raise public awareness would then be in control.
Its job is to raise public awareness. The other place I would put it front and centre is the Department of education because a lot of work has to be done with children as well. For instance, in Finland, their education system also teaches those who are engaging with the children - their teachers, families and whoever else. It builds confidence. It is happening a little bit in Ireland as well at a society level. I refer, for instance, to the Greystones smartphone-free project where schools work together. In a very well-resourced area parents themselves have been able to convince their children that all of them will stay off a particular smartphone. They will still have their social media, but not on a smartphone until they get to a certain stage in their lives. We could have public campaigns around that as well. Do we as a society want to stop children having access to these things? I recall that we had an event in the Children's Rights Alliance where an advocate from a very disadvantaged group of children in Ireland spoke about how, if their kids did not have access to their parent's phone in the evening, they would know nothing of what the other kids in school know. We have to build inclusive campaigns, so the one that may work for a very disadvantaged group is not the one we would be putting out on the 9 o'clock news for people who are still watching television.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
We will certainly look at the suggestion on Coimisiún na Meán and the Department. What about the role of the new national AI office?
Ms Noeline Blackwell:
Yes, we would have priorities for that. We would ask it to put children or child advocates onto the board in the first place in relation to that so that it is getting the information. That might be it. To go back to the Chair's point, however, if AI is a tool, people need to think beyond AI to the whole digital age in which people are living, so that they understand that, when they see the digital world, which is not that different from the offline world to an awful lot of young people, young adults included, public awareness is built around that as well.
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
I thank the witnesses very much for their input today and for their ongoing work in this space. Further to today, I might ask that the committee write to Coimisiún na Meán for an update with regard to those investigations on intimate images.
Sinéad Gibney (Dublin Rathdown, Social Democrats)
Link to this: Individually | In context | Oireachtas source
Could we ask both Comisúin na Meain and the Data Protection Commission?
Malcolm Byrne (Wicklow-Wexford, Fianna Fail)
Link to this: Individually | In context | Oireachtas source
Yes, sure. We will ask both of them. We might ask Coimisiún na Meán in terms of the information campaigns as well around people's rights. We might write to Assistant Commissioner Willis, who was quite useful. I appreciate there may be ongoing Garda investigations, but we will ask for a general update. We did specifically ask about the EU regulations with regard to CSAM and if is intended that they will come in by December. We might contact the Department of justice to see if it will be the case that they will be in place. Is that agreed? Agreed.
Other than that, I again thank everybody for being here. The meeting is now adjourned until 26 May.