Apologies have been received from Deputy Brian Brennan. Today's meeting has been convened to discuss and understand the approach taken by social media companies to limit and remove content that glorifies irresponsible and illegal activity, in particular content that potentially places the public in danger, along with the supervision and monitoring of tech companies regarding this type of content and whether the Broadcasting Act 2009 and the Online Media and Safety Act 2022 give Coimisiún na Meán sufficient powers to carry out its tasks.

From the Department of Culture, Communication and Sport, I welcome Ms Tríona Quill, assistant secretary in the media and broadcasting division, and Ms Sabha Green, principal officer in the digital policy and media plurality unit. From Coimisiún na Meán, I welcome Mr. Jeremy Godfrey, executive chairperson, and Ms Tanya Warren, director of policy. From TikTok, I welcome Ms Susan Moss, head of public policy and government relations, and Mr. Zachary Hecht, global head of policy. From Meta, I welcome Mr. Dualta Ó Broin, head of public policy for Ireland. From Google I welcome Mr. Ryan Meade, government affairs and public policy.

I want to point out that the e-commerce company X, formerly known as Twitter, was also invited but despite multiple requests refused to appear. As Chairperson I did not appreciate the manner in which it communicated that to our clerk, if I am being honest. It is very disappointing that a company of that scale refused to come before an Oireachtas hearing on these very important issues. I want to put it on record that it is not a good sign for a company like X to refuse to appear in front of a committee like this, especially considering the other invitees have decided to come. There is no reason for it to refuse to come before us.

I acknowledge the change of position of TikTok because it has a central role in what we are discussing. I also acknowledge that two representatives have been sent here. On 10 September I was invited to speak at the EU Presidency summit on online safety in Dublin Castle. Valiant Richey, the global head of partnerships, elections and market intelligence at TikTok, was on the same platform as me. As part of that event, I raised the fact that the company had refused to come in front of this committee. I acknowledge that there has been a change of heart and that because of that interaction a change of heart obviously was stimulated. Very early the following day we had communications from TikTok saying it now believed it should come in front of this committee, despite what it had said and the reasoning it had given previously. I welcome that. It is a good thing and it is good to see an organisation of that scale reflect and do the right thing and have a change of heart.

As for the format of this meeting, we have had a number of statements from each organisation. As Chairperson I am taking them as read, given the amount of time we have. We have all had time to digest them. There was no way they were going to be finished in three minutes anyway and we need the time to ask questions. This will be followed by questions and answers in the committee. The committee will publish the statements the witnesses have given us on our web page subsequently.

Before we move to today's discussions I would also like to clarify some limitations in relation to parliamentary privilege and the practice of the Houses as regards references that witnesses may make to other persons in their evidence. The evidence of witnesses physically present or who give evidence from within the parliamentary precincts is protected pursuant to the Constitution and statute by absolute privilege in respect of presentations they make to the committee. This means they have an absolute defence against any defamation action for anything they say at the meeting. However, they are expected not to abuse this privilege, and it is my duty as Chair to ensure it is not abused. Therefore, if their 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.

Members are reminded of the long-standing parliamentary practice to the effect that they should not comment on, criticise or make charges against a person outside the Houses or an official either by name or in such a way as to make him or her identifiable. We will now proceed with questions and answers with members. We are hoping to do two rounds of questions. The first round will be approximately five minutes for both the questions and answers.

I welcome our witnesses. Our thoughts are still with the families affected by the terrible incidents during the summer on the motorway. More generally, the challenges that we are all here to try to address are hugely vital. I do not like the title, but the EU Kids Act could be significant at a European level. We in the Green Party are looking at the issues around age verification tools, privacy proofing of those and generally trying to make a whole shift in the business model of our social media platforms away from user addiction. That is what I hope this committee is trying to tease out today.

In terms of questions, I agree with the opening statement from TikTok on the broader, cross-societal response, as that is important. There has been a lot of discourse on Garda youth diversion, the school supports, properly funding youth work and family wraparound services. We met Barnardos this morning and there is certainly agreement that what is happening on social media is symptomatic of a much broader societal challenge that there is an onus on us all to resolve. My 14-year-old son saw and was aware of the activities of some of the individuals involved. His peers are aware of it and there is the whole issue of the content being shared. I accept the points made by TikTok as well. It gave us statistics on the percentage of content that was removed before it was even reported and before there were any views. There is also a significant issue about content that was removed within 24 hours of publication. My question is really about that time gap before the removal of content. A little over 10% of content is removed within 24 hours but that 24 hours could be hugely damaging. Within an hour, content can go viral and be shared widely. The question is perhaps to all the social media providers on the response time to remove content in a timely manner. The figure of 90% sounds high but it should be higher and the response rate should be higher.

I thank the Senator for the question. First and foremost I want to note that this content has no place on TikTok.

Content promoting criminal behaviour and dangerous activities such as the theft of vehicles and dangerous driving is expressly prohibited in our community guidelines. We are constantly working to ensure that our moderation systems are highly effective and that we are proactively addressing and removing harmful content. No content moderation system is absolutely perfect across any online platform, just as enforcement offline will never proactively catch every bad actor but as the head of the trust and safety policy team, I and my colleagues across the trust and safety department are dedicated to continuously working to improve our systems. In light of the incident on 16 August, we have also dedicated additional resources to making sure that we are understanding the nuances of some of this content, which is highly contextual and requires additional analysis, review and iteration of our systems. We are dedicated to ensuring that the 90% number gets higher and higher over time through the use of additional automated moderation technologies as well as making sure that human expertise is at the centre of all of that work.

Are there supports in place for the teams who work on that moderation? It can be a hugely challenging area to work in because very often people are moderating content which is extremely disturbing, violent, sexually explicit and so on. Is that all handled and dealt with by staff in Ireland? What additional staffing resources are there? Mr. Hecht suggested that some additional AI supports would be put in place but in terms of the human resources involved, that work takes its toll on the people involved. What supports are there for the people working on that?

I thank the Senator for the question. Speaking generally about our global moderation workforce in front-line moderation, we make sure that they have support on the ground as well as specific modules to address any potential psychological harm they might be experiencing through the moderation of content. That being said, as we increasingly leverage automated moderation tools, we are aiming to reduce the amount of highly graphic, explicit and harmful content that humans have to review. One of the benefits of automated moderation tools and technologies and their increased capability is that we can shift that highly graphic, clearly harmful and clearly violative content out of human moderation, ensure that human expertise and human moderation are reserved for the more nuanced, contextual types of decisions and lower that potential psychological burden on front-line moderators.

I am over time but are there supports in place for the staff who are working on moderating that content?

Yes.

I am referring to psychological supports.

Yes.

I welcome all of the witnesses here this evening. I will go straight to questions because we only have five minutes. The dangerous driving issue aside, I wish to briefly relate my own experience. My five-year-old daughter, who I will use as an example, has access to YouTube or whatever for a specified period every day. She would have seen content online about the cat in the hat phenomenon that has been spoken about widely. Whether children are seeing it or not, other children have seen it and are talking about it to the extent that in my own circumstance, my five-year old daughter is now terrified to go down to the green with her friends because of this cat in the hat phenomenon that was prevalent a number of months ago on social media. This has a profound impact on people.

The dangerous driving one is obviously the one that has caught an awful lot of the headlines. I agree with that no system will be perfect. I assume that the platforms are predominantly using AI tools to police this and the human element of it probably comes into the investigative piece at the end. No system is perfect. I accept that but it needs to be pointed out at the start to the three companies present that the data they have presented to us today seems to be global. Will they talk us through, starting with Google, the number of Irish users who were exposed to the dangerous driving video that triggered this discussion? How long was it up there? How many people viewed it in that time? I ask the witnesses to briefly speak to that.

I set out in my opening statement the range of protections we have in place to try to ensure this type of content does not have a place on YouTube. That goes back to the policy explicitly calling out dangerous driving challenges, where content is being prohibited. It comes through the enforcement that we do of that policy, which is proactive and targeted. It comes through the use of our intelligence desk and our intelligence partners to ensure that we are aware of trends that are emerging and also the built-in protections we have on things like livestreaming. With all of those protections in place, we really saw a very low volume of content related to dangerous driving.

I want to quantify that. What does low volume mean?

Less than ten videos and less than ten channels were removed-----

In what period?

In the period from the start of July to the end of August.

I will put the same question on the specific volume of Irish dangerous driving videos to the representatives of the other two platforms.

I am going to give the Deputy two sets of figures. One is related to the dangerous driving policy and the other to the viral challenge policy and the impact of our designation of this content - driving against oncoming traffic - as a viral challenge. In the period from 1 December 2025 to 5 September, we took enforcement action against ten items that were Ireland originated. That could be videos from elsewhere in the world but they were uploaded by an Irish user. Eight of those were age-gated and two were removed. Age-gated means that they are not accessible to people under the age of 18. In light of the events, we took a decision to designate driving against oncoming traffic as a high-risk, viral challenge. The result of that is enforcement action on 130 items of Ireland-originated viral challenge content. Of those, 127 were removed and three were age-gated.

I think I know the answer already but on the age part, what steps does Meta take to ensure that under-age people are not viewing that content?

Our approach to age is multifaceted. We take a number of steps, including checking patterns of behaviour to ensure that they are the age they say they are. We recognise that age is a challenge and we have been calling for legislation at the European level for quite some time. We take a whole range of steps in relation to age management but we realise that it is not a complete picture and obviously the KIDS Act is going to step into that as well-----

I do not want to be rude but I want to get TikTok to provide the numbers for Irish-origin videos in answer to the same question. If I am a child and I have clicked on one dangerous driving video, how likely is it that I would receive more and more suggested content like that? Maybe they could explain how this algorithm that everybody talks about works. I ask all of the witnesses to answer that question as well.

We do not classify dangerous driving within our data separately from our overarching category of dangerous activity and challenges. We have the data for each one for that overarching category but that also includes things beyond dangerous driving like risky behaviour at certain heights or the use of tools in an inappropriate and dangerous manner.

I am sorry but my time is nearly up. If that dangerous activity is classified as one cohort, how many dangerous activity alerts were on TikTok's system during the same period? I am not saying that is representative of what we are speaking about but I just want Mr. Hecht to give us an indication.

Certainly. It is a broader category, not specifically Irish driving, which is a smaller part of that category. For H1 of 2026 it was approximately 42,000 pieces of content. Putting that into context, on a daily basis there are around 400,000 pieces of content posted on TikTok, so that 42,000 number is over H1.

The issue I have is that Mr. Hecht is saying that it is smaller than the 42,000 number he just referenced but for him to say it is smaller, he must have an idea of how many incidents there were related to dangerous driving.

In the time since the original incident we have been conducting additional sweeps and analysis on the platform. It is not representative of the H1 data that I shared but in that period, over the previous month, we have only identified approximately 1,600 instances of content related to the broader category of vehicular theft, highly risky driving and behaviours related to that. It is a smaller proportion of that overall number, relatively.

TikTok does not have Irish figures.

That is the Irish data. It is just not necessarily specific to the high-risk driving category but is part of a larger subset that we have.

Why can Mr. Hecht not get those figures?

What we are focused on is removing content and making sure that we do not have that specific-----

I will come back to that. I do not think that is really an answer.

I will start with the conversation that Deputies had about removing content and direct my first question to Mr. Hecht. We can all agree that AI is very intelligent. Before videos are uploaded and before we have a conversation about removing content, how quickly it can be removed and how much of it is removed, why can it not be curtailed or prevented from being uploaded in the first place?

We know how intelligent AI is and if you put an image into ChatGPT or Gemini or any of these AI generators, many of them will come back to say, "No, I cannot do that because that is a minor." Why cannot TikTok spend money on the AI that will refuse to upload these videos in the first place?

I thank the Senator for the question. We are making continued investments in improving and advancing our automated detection systems, including the use of AI. At the point of upload onto TikTok, every single video is moderated by that system. We are increasingly removing content through those measures, so-----

What does Mr. Hecht mean when he says it is moderated? It is all getting through.

No content moderation system is 100% perfect, but we are already addressing over 99% of videos before there are any reports and over 83% before there are any views in this larger category of dangerous driving and dangerous activity.

I saw those percentages. Does the "dangerous activity and challenges" category that Mr. Hecht referenced include representations of gender-based violence? The last time we were talking it was about pornographic images and deepfakes. This time it is dangerous driving. I am conscious that this is very triggering for the families involved but I wonder what TikTok means by criminal activity. Does that include representations of gender-based violence?

Gender-based violence is prohibited in our community guidelines. It is a separate category than the one I am referencing.

It is not included in these percentages.

It is not.

What are these percentages then? It is not clear how many videos we are talking about. Do these figures just relate to dangerous driving?

The information provided in the statement is related to the category of "dangerous activity and challenges", of which high-risk driving and risky driving is a subset.

It does not include everything that is dangerous or everything that is criminal.

We have other policies that address types of criminal behaviour around theft. We have policies that address hate speech, bullying and exploitation. Those are separate categories within our community guidelines. We publish our overall community guidelines data and I am happy to go back to the team and get data for the committee on those other areas.

It would be good. It sounds like TikTok has an awful lot of areas that it needs to police. Mr. Hecht has listed off another four or five areas that it has to police. Having to remove these videos must take up an awful lot of time. Could TikTok not just invest its millions in more intelligent artificial intelligence so that these videos could not be uploaded in the first place?

We are making significant investments in that space. Addressing the entirety of our community guidelines is something that I, my team and the thousands of trust and safety professionals around the world are dedicated to doing. On a day in, day out basis, we are focused on improving that and driving up our numbers. I note that our removal before any reports in this category and broadly is around 99%. I am happy to get back to the committee with specific data on other categories. We are continuing to invest and to leverage increases and improvements in technology to do this at scale and better.

The UN released some figures. Mr. Hecht referred to a 90% figure. According to Sensity AI, 90% to 95% of all online deepfakes are non-consensual pornographic images of women. These things should never have been uploaded in the first place.

Is Mr. Hecht aware of a report by Dublin City University, DCU, and the Department of justice, "Recommending Toxicity"? He probably is. It is all about TikTok.

I am not sure that I am familiar with that specific report. I would have to take a look.

I am happy to send it to Mr. Hecht. It is quite galling. The researchers set up a load of accounts pretending to be 16-year-old boys and 18-year-old men. Some of the accounts went and looked for male supremacy and violent, anti-feminist videos and some of them did not, and then there were control accounts. All ten accounts that were set up were sent to the same videos. They were sent to the same representations of gender-based violence, anti-feminism and male supremacy without even seeking it out. That brings us again to the recommender systems. We have talked about the issue previously at this committee. Where is TikTok with turning off its recommender systems and keeping track of these algorithms? It is failing.

Content that is violative, such as what the Senator has just mentioned, and I would need to take a look at the report to be more precise, should not be on the platform and that means it should not be recommended. We are dedicated to removing it and getting better. I also note that certain experiments or different approaches to trying to be a user on the platform do not necessarily reflect the entirety of or the common experience of users. I would want to take a look at the report to ensure I have a full understanding of what it is.

It was funded by the Department of justice. I will send it to Mr. Hecht later. He has just said that such content should not be on the platform. TikTok needs to do better at not letting it be uploaded in the first place.

We are constantly working to improve our systems. Like any online platform, it is not a 100% effective system. However, we are increasingly getting better at proactively removing harmful content online. That is what my team and my entire department are dedicated to.

Time is short, so I will plough straight on. Sadly, I had to come off TikTok. It is not a place I would want to be and it is certainly not a place I would want my children to be. Mr. Hecht mentioned that no AI is perfect. That is not good enough. If that is the case, humans have got to be in place until that AI is perfect. As my colleague said, if that much content needs to be removed, perhaps the concept needs to be looked at again because it is not exactly fit for purpose at the moment.

I am going to concentrate on TikTok and the percentages that Mr. Hecht has given. He said that 83.4% of illegal videos removed had no views, which means that 16.6% were viewed, while 10.1% took more than 24 hours to remove. I have simple questions. How many videos are we talking about? How many views did they accumulate? How long did the most serious cases remain online? I am not looking for percentages; I am looking for figures.

I want to clarify that this category does not necessarily include illegal content. This is content that we prohibit in our community guidelines. We go beyond what may be illegal.

It is content that TikTok would not want the public to see-----

It is content-----

-----which again takes in illegal activities.

It is content that is violative of our community guidelines that we want to ensure we are protecting users from and dedicating ourselves to removing.

How many of those videos do people see?

In terms of the data that the Senator is noting, in the first half of 2021 there were 42,000 pieces of content. That is content that ranges from behaviour that is potentially risky at certain heights to inappropriate use of specific tools that could be dangerous. It is a very broad category, which-----

That is what Mr. Hecht said to the Deputy.

-----dictates a very advanced approach.

Mr. Hecht said that. How many views do they accumulate? I am not talking about percentages or numbers. Those videos could have been viewed millions of times. How many millions of people saw this content that they should not be seeing?

I do not have the information on the specific views to hand, but it is something I can go back to the team to see if we can pull.

I would like those figures and figures for how long the videos remained online so people were still allowed to see them.

I will again quote the percentage for how long they remain online. Some 83.4% were removed before-----

I do not want percentages. I feel that Mr. Hecht should have come here with the numbers of people who saw this content. I would like those figures, if possible.

I will move on to something-----

It is deeply concerning that Mr. Hecht does not have quantitative figures.

It is.

TikTok had enough notice. All of the organisations had enough notice. It is concerning. I do not think representatives should appear in front of a committee without the figures. Mr. Hecht might want to ask somebody to get those figures while we are sitting here.

Go raibh maith agat, a Chathaoirligh. That is exactly what I mean. I need to know about people and the number of eyes that saw this and how far this content spread.

I will move to YouTube. I sat beside Declan McNally at an event in Northern Ireland last week. Stephen McCullagh used a fake six-hour YouTube livestream as an alibi while he murdered Declan's sister Natalie in 2023. It remained on YouTube for more than three years and accumulated over 750,000 views. Three quarters of a million people watched the video that he used as an alibi while he murdered his partner. This caused untold upset to the family, as I heard from Declan only last week. Why was it left online for so long until his conviction?

I must admit that I am not familiar with the specifics of that case. I am aware of the case and the fact that part of his alibi involved his role as a YouTube streamer. I will have to check with the team on that particular incident because I do not cover Northern Ireland. It is just one of those things. I have a colleague who would probably be better able to-----

It is still YouTube and it is still-----

I understand.

Why would Mr. Meade imagine that was the case? I know he is probably not qualified to answer that but I want to know the answer to these questions.

Sure.

I would like them to be brought. It was a part of a criminal investigation. Stephen McCullagh's account was closed and demonetised after his arrest. YouTube knew enough to know that it was criminal activity or party to a criminal act. It knew enough to demonetise his account but not remove the video.

I am not familiar with what is in the video. The first thing would be that our teams would review it against our policies. As I understand it, the video was not something referring to the crime. It was simply something that was, as the Senator said, part of the overall case.

YouTube knew enough to remove his account.

I would have to check what the procedure would be in that case.

In general, if the video does not violate a policy, there may be no reason to remove it. One should bear in mind that under the Digital Services Act we now have obligations in both directions. We have obligations to report on the reasons for removing as well as the reasons for not removing. I will check.

It is a clear example, though, of something that needs to be checked because sometimes illegal activity or something connected to illegal activity can remain for up to three years. I will move on.

I am happy to check that and come back to the Senator.

If something is deemed illegal activity, what happens after it is deemed that? Is it reported to the Garda, are there follow-ups and is a notice sent to the poster saying, "You have put up illegal content and it has been reported to the Garda"? Any of the platforms could answer that question, I presume, possibly starting with TikTok.

I thank the Senator for the question. Typically, we report content where there is an immediate threat to life or an immediate threat to harm. That is escalated immediately to An Garda Síochána. It can sometimes be within minutes. We try to make sure it is within less than an hour to two hours. We have a dedicated law enforcement team based here in Dublin made up of former members of An Garda Síochána and they have those relationships and that connection with An Garda Síochána.

What do any of the other platforms have to say about cases in which it is not that someone is in immediate harm but it is deemed illegal activity?

Similarly, as part of our response to incident responses or crisis response, we would have several teams involved, including our law enforcement outreach team. They, similarly, would have a very close relationship with An Garda Síochána. An Garda Síochána can request data from us when it is investigating a crime. If certain thresholds are met - it is difficult to be completely black and white on when those thresholds are met - there would be a notification to An Garda Síochána.

I probably do not have time, and might leave it until the second round, but could we have the figures of how many reports have been made in criminal illegal activity or those who are unsafe?

I thank the Senator. Before I call Deputy Cleere, who is substituting for Deputy Malcolm Byrne, I want to say something to all of the e-commerce companies here about statistics. If they are asked for statistics on the quantity of views - it was quite obvious we were going to ask about such matters - they might not have them. In relation to TikTok, it was 16.1%. They referred to 42,000 videos. That sounds very small, but they do not know the number of views. Can they contact somebody now to find out the number of views, and get that figure while we are in here in committee? Referring back just does not work. They should know the number here. Will they please get that data now, while they are sitting here? They should email to ask someone to get it before this meeting ends. They are here in front of the members of the committee. Before they leave the committee, we want to know how many people in Ireland viewed those videos.

I welcome all the witnesses and thank them for their comprehensive presentations. I have spoken to dozens of people over the past number of weeks who are afraid of their life to drive up or down a motorway as a direct result of what happened on the M9. I want to be specific and clear in my questioning. My first question is to TikTok. Exactly how many dangerous driving videos has TikTok removed in Ireland to date? They say in their opening statement that they had thousands of dangerous activity videos in Ireland in the first half of 2026. How many is that? How many videos have they removed from their platform?

For our category of dangerous activity and challenges, which includes dangerous driving but is not exclusively dangerous driving as it includes other areas such as certain behaviours that are risky at certain heights, as I have mentioned, that number was 42,000 in the first half of the year.

Forty-two thousand?

Forty-two thousand videos. There are approximately 400,000 videos posted a day from users in Ireland on TikTok. Not all of that 42,000 is dangerous driving behaviour. In the time we have been monitoring since 17 August in relation to the incident, we have identified only - relatively - 1,600 pieces of content that are related to areas such as the theft of vehicles and risky driving. This, importantly, is not specifically wrong-way driving. It is, more broadly, specific types of speeding that might be over what we consider to be appropriate, as well as specific manoeuvres around vehicles on roads, but it is not also specifically wrong-way driving. Out of that 1,600, under 2% has any indication of wrong-way driving and it is not highly-----

So the number is 32.

It is under that number. It is pulling the specific data, but that is around where it is.

"Less than 30," is basically what Mr. Hecht is saying. That is fine.

The thing that is important is that it is not always readily apparent because there is no other car on the road.

What Mr. Hecht is basically saying is that approximately 30 videos were taken down by TikTok.

The number of videos related to issues of wrong-way driving is under that. There is not always-----

In terms of wrong-way driving, I have seen way over 30 videos from different people. Is there still content out there?

One thing that I also want to note is that-----

I am conscious of time now. I do not mean to rush Mr. Hecht but, from TikTok's perspective, is there still content out there of wrong-way driving?

One of the things that I want to note is that a depiction of wrong-way driving is not absolutely going to be removed when it is done to raise awareness of the issue. We see content from people and communities that want to raise awareness and share information about harmful behaviour that might be happening offline. It is very important that we provide people with the ability to express what is going on.

The Chair referenced the absence of specific data on the number of views of the videos that have been taken down, etc. I saw some that were probably in the tens of thousands and the hundreds of thousands. Hopefully, we will get that figure by the end of the meeting. I want to go a level deeper, if I can, with reference to the amplification of a video when it goes up. In this case, it is a video of wrong-way driving. It was said in an opening statement that "86.5% of the videos ... removed had ten views or fewer." That is very reassuring, but what about the other 13.5%? Do the witnesses know how many views they had? What is the highest number of views that a dangerous driving video, which would have been removed by the companies under the policy, would have had before it was taken down? Surely Mr. Hecht knows. If there was only 30 of them, surely to God he knows what the most views one of them had was.

That is not something I have at hand but-----

Come on, lads.

-----again, we will get it.

There is only 30 that have been taken down and Mr. Hecht does not know what the highest number was.

It is not something that I have at hand right now but it is something that we can look into and get back.

That is not very professional. Of the videos that received a significant number of views before removal, how many - this is really important - were recommended by TikTok's algorithm to people who had not searched for or followed that type of content?

That is also not something I have at hand, in terms of how that content was accessed. We have the number of videos that we are discussing and we can go back to the team and see what is possible to pull on this.

TikTok used the word "amplify" in its opening statement, but Mr. Hecht cannot tell this committee how many of these dangerous driving videos were amplified by its recommendation system before they were removed. Mr. Hecht cannot provide that data to the committee here. I do not think that is not acceptable.

I am conscious of time. I am looking up at Coimisiún na Meán here and I have a question for it. Is Coimisiún na Meán 100% happy when it looks to the companies here that they are doing everything within their power to take down this harmful illegal content quickly? If Coimisiún na Meán is not happy, what is it doing about it or, more importantly, what will it do about it?

I thank the Deputy for the question. There is a number of regulations.

The first question was this: is Coimisiún na Meán 100% happy that the group of companies here are responding 100% to the best of their ability?

A regulator will never say whether they are happy.

Mr. Godfrey is failing to say he is 100% confident that they are doing everything they need to be doing.

If I might say, we are engaging-----

It is very worrying that Coimisiún na Meán, the regulator, is failing to endorse the companies here. Am I correct in saying that Coimisiún na Meán does not have confidence in them?

Our role is to assess what the companies are doing. If we have concerns, we can take further steps such as investigations. We are engaged in an assessment of the situation and, because it is a regulatory process, we cannot give any details of that until we reach a conclusion.

I am not looking for specific details. In general, is Coimisiún na Meán 100% confident with the companies here: Meta, TikTok and Google?

It is our job to look at what they are doing and to identify-----

Mr. Godfrey has had three chances now to voice confidence in them.

To be honest, I would not express confidence or a lack of confidence in companies. We are a regulator. We have a legal process to follow. Our role is to assess whether regulated entities are complying with their obligations and take actions-----

Are they complying with their obligations?

As I said, we are carrying out an assessment. We have opened a number of investigations where we think that regulated entities are not complying with certain of their obligations-----

How many?

-----such as, in the case of Meta, WhatsApp and Instagram, whether they are providing people with a real choice.

I am not asking for specifics.

I call Deputy Joanna Byrne.

This is my last question. How many cases have there been?

We have six formal investigations that we are involved in on our own and four that we are doing jointly with the European Commission.

I thank all the witnesses for being here. If we were to take these opening statements at face value, we could be led to believe that there is no problem with illegal and unsafe online content. Neither the Department nor Coimisiún na Meán mentioned any of the scandals this year in their opening statements, be that the Grok AI nudification scandal or the teenagers doing TikTok challenges that lead to deaths and serious injuries on our roads. Herein lies the problem. We have a hands-off regulator and a Department led by a Minister who said that he does not believe social media companies should be fined.

I will start with Coimisiún na Meán. The opening statement that it "ensures that platforms follow these rules, and when they do not, we take action. We have six open investigations into potential breaches of online safety rules, and we are collaborating with the European Commission on another four". In all seriousness, do the witnesses think that this constitutes taking action? Having listened to the contributions, there are more than ten circumstances that need investigation, never mind some of the figures we have from some of the social media companies alluding to thousands and hundreds of instances, and Coimisiún na Meán is dealing with ten cases. Seriously.

We are a systemic regulator. We have cases involving breaches of obligations to provide users with a choice of recommender systems, including one that is based on profiling. We have a case about appeals against moderation decisions. We have a case about age assurance. Jointly with the European Commission, there are cases about the recommender systems, which the members have specifically spoken about, and the addictive design features. We are doing cases jointly with the European Commission about those aspects. The European Commission has made preliminary findings on all those instances.

We are talking about ten cases. There are more than that in respect of the lives lost in the TikTok issue alone and the content around dangerous driving and wrong-way driving on our motorways.

Yes.

I emailed Coimisiún na Meán multiple times this year. I emailed in January and in August urging it to take action against X over the Grok scandal. I only got a reply in September. It took Coimisiún na Meán nine months to open the Grok investigation, and no fines have been issued. Is this the way Coimisiún na Meán is going to operate in future? Let us be transparent here. Has it been instructed to operate this way or is it the Coimisiún na Meán's own choice? Why is Irish legislation not being used? There seems to be this stand-off approach of waiting to see what Europe does. European legislation should be an aid to the strong Irish legislation that is there to tackle social media companies when they breach these regulations. Is there a fear, a culture of fear, of taking on social media companies in Coimisiún na Meán?

We are absolutely independent and fearless. We will-----

It does not come across as fearless, to be fair.

If we see non-compliance or if we are concerned about non-compliance, we will make sure that we use the most appropriate tool to address it. In the context of Grok, what was identified as the most appropriate tool and the regulatory obligation that was engaged was the obligation to assess and mitigate risks around non-consensual intimate image generation and distribution. That is something the European Commission leads on. We are working with it. The investigation was opened within weeks of that issue coming up.

It is really important to us that we follow the law and do things appropriately. It is a very litigious environment. If we do not follow the processes that we have to follow, we will just get challenged in court and we will not be successful. I can absolutely assure the Deputy that we are extremely concerned about a number of different harms online, whether it concerns children, gender-based violence, terrorism or-----

The commission is just slow to act, maybe.

No. I think-----

I am going to move on to the Department. The Department's opening statement states:

With respect to online content, the Commission is expected to come forward with a legislative proposal during Ireland's Presidency to amend the audiovisual media services directive, including in respect of video-sharing platforms. The Irish EU Presidency is committed to giving priority to the negotiations on these proposals.

Can Ms Quill give us details of what these amendments are and how the need for them was identified? Was it an oversight during the work done by the Future of Media Commission, which published its own recommendations in 2022, not to include this at that time?

The revision of the audiovisual media services directive, AVMSD, is being undertaken by the European Commission. It has said that it will bring forward proposals in the course of this year, so we are expecting them in October or November. The revision of the directive is in recognition that more needs to be done on the protection of minors. It is being done through a number of different pieces of legislation, including the EU KIDS Act. We would expect that the revision of the AVMSD will likely include more detailed on specific categories of harmful content, which I think will be useful at a European level in future.

Will the Department be able to keep the committee informed of that as it progresses?

Yes. We would be very happy to brief the committee.

I will come back in during the next round of questions.

Picking up on what Senator Noonan mentioned in respect of TikTok's opening statement around the broader societal context, I disagree. While I recognise the broader societal context, I believe these social media companies are being let off the hook. They are self-regulating. They have community guidelines, content moderation systems and significant investment, but none of it is working. We are listing off a gazillion examples of how TikTok's technology is harming us, our children, our populations and our societies. One of the most prescient ones is that a country manager in trust and safety is not even aware of the details of it. None of this is working.

I want to broaden that from the reason we are here today and talk about some of the other flashpoints, things like, for example, the Dublin riots, the Belfast riots, some of what is happening in the UK, in Portsmouth, and even what happened with the Rohingya population in Myanmar. These are all examples of times when platforms have been used to whip up violence. The difference between violence online and in the real world was spoken about, but some of this is being sown on these social media platforms. It is growing on these companies' platforms. Nudification apps are online-created harms. It is not, therefore, about finding out how they are dealt with online but about addressing what are harmful products. These social media companies' platforms are harmful. While most content might be caught before it garners much attention, there is a reality that those videos fanned the flames of those incidents in Belfast and Dublin. Before the platforms hit pause, the algorithms have already detected that it was just the right combination of hate and rage to get the companies profit, and that is where we end up.

Outside of those sudden flashpoints, I want to talk about one specific case, which happened just a few weeks ago in Fagersta in Sweden. A 17-year-old man, Liam Nebel, walked into a school armed with a sword and stabbed a 17-year-old female student to death and wounded three others. He had become very popular on TikTok, particularly in the "true crime" community, for his edits praising school shooters and attackers like Anders Breivik. Almost immediately, similar videos appeared of him. This is the glorification and lionising of hateful content and actual murder.

I ask Mr. Hecht to please not give me community guidelines but to tell me how TikTok's algorithm interact with trust and safety. As the head of trust and safety, how does he interact with the designers of that algorithm and the C-suite executives who decide the difference between profit and trust and safety? How does that happen? How are those conversations happening?

We do not seek to incentivise and we remove content that violates our community guidelines. That includes content that might glorify or promote criminal actors, violent actors or violent behaviours. We do not-----

TikTok does not seek to incentivise, but does it recognise that its profit model does?

We want TikTok to be a platform for-----

Does Mr. Hecht recognise that TikTok's profit model does so?

Let Mr. Hecht answer.

Our model is to ensure that users come to TikTok for a safe environment where they can experience community, joy and creativity. Trust and safety is fundamentally at the heart of that.

Does Mr. Hecht deny that extreme content is more profitable for his platform?

Yes, I would deny that. TikTok-----

Mr. Hecht denies that. He denies that radicalisation and polarisation are profitable for TikTok. He is saying they are not profitable for the company.

We want TikTok to be a safe and inclusive platform for all users. Trust and safety, in our department and in our work, is at the heart of making TikTok a place where users want to be, want to come, want to share and want to view content. We take our role in upholding that for the platform extremely seriously.

Will Mr. Hecht give me an example? In this case, in Sweden, how does that conversation go? Why is this content still there? It is still there. It is clearly against TikTok's guidelines but it is still being shown to the victims and survivors of this exact incident. That content is being shown to them now. Has Mr. Hecht had conversations with his c-suite executives to say that they have any issue where trust and safety are being breached because the profit model is making money from it?

Safety is my top priority and it is a top priority for the company.

Has he had conversations about this specific case with his colleagues?

This specific case and the larger true crime community is something that my colleagues have worked on. I am focused on-----

Has Mr. Hecht had conversations with his colleagues at c-suite level about this specific incident?

I have been speaking with my-----

C-suite is corporate suite. It is the CEO and relevant heads of departments within a tech company.

I have been working with the trust and safety team to make sure that we are improving our systems.

Mr. Hecht has not discussed this with, for example, the chief executive of TikTok or any of the senior people on his peer level who are not in trust and safety.

I personally have discussed this with the head of the trust and safety department, in terms of the broader topic.

Have you discussed it with anybody outside of trust and safety?

Discussions about user safety and trust and safety are regularly happening with the leadership of the company. We continue to invest and focus at leadership level on improving our moderation systems and continuing to invest in advanced technology to make sure content that violates our community guidelines is being removed. We are working on getting that number higher.

Can I confirm that with Mr. Hecht's concerns about the true crime community and the proliferation of crimes happening in people's lives, and the fact that people are dying, he has not discussed that with the head of engineering in TikTok, the head of marketing, the head of product, customer services or any of those people? Has he had discussions?

The leadership team has regular discussions about how to address harmful content across all of our prohibited areas. It is a core focus of the company and of our entire leadership team.

You have not discussed this specific item.

I have not had a discussion about the specific item the Deputy is referencing, but on a regular and frequent basis we are discussing how to keep this platform safe, and how to continue investing in advanced technology as well as human expertise to make sure that we are continuously increasing the amount of harmful content we are removing, proactively and before anybody views it.

It is not working.

My first question is for Google and Meta. How much did Google spend in 2025 on online safety and that whole category?

We do not break out those figures.

What does Mr. Meade mean Google does not break out those figures?

It is just the case that we do not-----

Mr. Meade does not know how much was spent on online safety. I thought this was a pretty easy question.

I do not have a figure, but we have various figures we publish about the number of content moderators we apply across different parts.

You do not know.

I do not have a financial figure.

Next.

I do not have that figure. I have the figure of views if that is of any interest, and the Deputy wants to do that now.

I will do that later. TikTok knows what figure it spent in 2025.

I do not have a specific figure but in our-----

Hold on a second. At a conference on 10 September, Mr. Valiant Richey, who I thought was very good, gave a figure.

Previously we have indicated we spent $2 billion per year on trust and safety but in our Irish office our people are our largest expenditure, and three quarters of the people in our Irish office are dedicated just to the role of safety at TikTok.

He has publicly said $2 billion per year.

It could be more. That was the last figure-----

It was the figure that startled me at the conference and it startles me now. It can be imagined what profits are being made if TikTok can spend $2 billion per year on online safety. Dare I say it, after what we have seen in Ireland in the past number of months, it is not very well-spent. We are at a watershed moment in this country. We are at a watershed moment for these companies, not just TikTok, but it is the platform of choice for most young people as we all know. That figure is startling. TikTok can afford to spend €2 billion on online safety and we can still on a daily level have my 16-year-old daughter, with whom I saw those videos, able to show them to me. She was able to tell me those videos were still there, when TikTok is spending $2 billion on online safety. She is 16 years of age. No one should see those videos, let alone a 16-year-old. I know you can put that out as an isolated case but it is not. I have seen loads of this stuff, and across a range of other categories. The number of videos companies say they have taken down, versus the number anecdotally we have seen collectively, shows that those cannot be right.

We have a serious problem here. We cannot continue. Personally, I do not believe the EU is going to get the point where it can deal with this so Ireland, as a country, is going to have to deal with it. We have to protect our citizens and protect our children. That basically means we have to bring in legislation and regulations. The companies are going to have to adhere to the regulations. The whole dial is going to have to change. Instead of countries trying to meet the companies' requirements, they are going to have to meet our requirements. If they do not meet those requirements over a period of time, based on an audit or on a percentage of failings, they will have to be held criminally liable. When they set foot in the EU, they will have to come to Ireland to answer questions. That is where this has to go because this is dangerous and illegal content. RTÉ, Virgin or any of the national newspapers could not do any of this. It all comes down to the bottom line of these companies not being publishers. That has to end. We accept that no moderation is perfect. It is impossible. It is utopian but there has to be legislation and regulations, and they have to have the necessary backbone where people will be held criminally liable because fines do not work. An e-commerce company was recently fined and their share price went up. It is never going to work because it is built into the baseline. The dial has changed. We have had a watershed period and as a country we are absolutely going to have to stand up for our people. I do not think it is acceptable any more. I do not believe we can continue in the same way as before.

I have a question about the data I requested from TikTok. The figure is 42,000. My colleagues asked for the number of views. We all want to know this because it would show the impact on the Irish public. How many people in the Republic of Ireland - or on the island of Ireland in case there are geocoding issues - actually viewed those 42,000 videos?

We have asked the team. I do not have the data that was requested. The team is taking a look at it and we commit to seeing what we can get for the committee. I can say that 83% of that content was removed at zero views. I can also share, and the team has sent over, that 96% was removed at between zero and 100 views. We break out the data-----

What we really want to know is the 16.1%. How many views? A certain percentage was between zero and ten and a considerable percentage was over that. We want to know how many views. Is that in the hundreds or in the thousands? I suspect it is in the hundreds of thousands or millions. That figure is the figure that shows what I said earlier, which is that it cannot continue.

That is not data I have right now. I am speaking with the team and we will see what we can pull. We classify the data in a slightly different way but I understand what the Deputy is asking. We are focused on making sure our number of removing approximately 83% of content before anybody has viewed it can be as high as possible. That is why we are continuing to invest in technologies. I understand the concerns that have been shared. My team, the entire department and the company are focused on this matter. We are leveraging those advanced technologies and our increasing knowledge of the space to do better and better.

I accept Mr. Hecht's bona fides but here is my comment: I believe legislation should be brought in. I believe the regulations with Coimisiún na Meán should probably be reinforced, based on the experience of my colleagues and I. It should be that if a company misses a certain percentage over a period of time it becomes liable. First, the content should be taken down and, second, the people held liable. We cannot have the people of our country being treated this way. The whole dial has to move and it has to change. It is only when we do that, given the zillions these companies are making, that there will be change. This has gone too far in the wrong direction and states and legislators have to pull this back.

To add to that point, content that depicts a crime is not necessarily illegal. This is the point the Cathaoirleach has made. Under the EU legal framework, the platforms are not liable for illegal content as long as it is removed in a timely manner. This goes back to the point I made at the outset, which is that the 10% slipping through the net could have been up for 24 hours doing huge damage. I also do not want to discount the points I made earlier around the investment in youth work. It is not unrelated to what happened on 16 August, and it is important to say that. Perhaps the coimisiún could answer that question.

Going back to the Chair's point, is the sharing of the content something we should be legislating for? It is quite shocking to think it is not illegal while it is up there. Is that a cop-out for the social media companies to say they have the luxury of a little bit of grace and time to remove it when they have become aware of it?

To clarify, one of the things we said in the opening statement is that you will get some depictions of crime. For example, you might have a news report of a public order incident that will depict crime but that is not glorifying or encouraging crime. Not all depictions of crime fall within what we are concerned about. Obviously, as an agency we are also very concerned with freedom of expression and the balance of freedom of expression and the protection of the public. There are certain types of content that depict crime but do not glorify or encourage it and are actually legitimate exercises of freedom of expression for news reporting or civic discourse.

To answer the Senator's other question, the current legal framework is that once companies become aware of illegal content, they must remove it if it is manifestly illegal. If they know it is illegal and it is clear that it is illegal without having to go to lawyers to check, then either they must remove it as soon as they can or they become liable for it. They become a publisher, effectively. The Chair talked about them not being a publisher. They become a publisher once they have actual knowledge of the content.

Is it difficult to prove they have become aware of it?

The legal framework of the Digital Services Act provides for a notice mechanism. Law enforcement or any other user can inform the platforms about illegal content. The moment they are informed about it, they are deemed to be aware of it. They also, as members have heard from the companies, do proactive monitoring on their own, using both humans and AI. That is something we welcome. We want them to be as effective as possible when they do that. They also have an obligation to make sure they assess and mitigate risks to public safety, public security and the health and well-being of individuals. The way in which they define their rules and enforce them are obviously important mitigation measures where we have concerns around that. That is a European Commission competence, but we work with the Commission on that. That applies not just to the companies based in Ireland and which are in this room, but also to companies based elsewhere in the EU. For example, Snapchat is also very popular with children across the EU, including children in Ireland. The European framework provides us with a mechanism for ensuring companies not based in Ireland are also held to account for protection.

I have a second quick question for the platforms. If AI is to be the great technology we are constantly being told it is, why can we not get to 100%? Why can we not get to a point where the content Deputy Gibney spoke about can be shut down immediately, and not make it onto the platform at all? If the parameters of the community safety guidelines could be tightened to such a degree that AI and human moderation could both interact to ensure we get to 100%, should 100% be the default so that this content never sees the light of day?

I am happy to take the question. The reality is that content can be highly contextual and ambiguous. Two humans can look at a piece of content and come to different conclusions despite having hours to look at it. There is an ambiguity in a lot of content. We hope and expect that with the advances and increases in the ability of automated tools, we can continue to address content and remove more and more. However, there is always going to be ambiguity, both because humans might not agree on it and because signals might not be readily apparent in that content.

Taking some of the content we are discussing around driving, it is not always immediately evident what is going on until you have that contextual information about what is going on offline. We are continuously working to embed additional signals into our systems, but there will always be an element of ambiguity in addition to there being a great deal of complexity in scaling those automated decision systems.

Picking up on what Senator Noonan said, I am not sure Mr. Hecht has explained why platforms cannot just have a block on these videos and images being uploaded in the first place, because AI is so intelligent in every other way. It is the first question I asked in the first round. I am not sure, given the resources TikTok has globally, how making sure the videos do not go up in the first place is not the first port of call . Mr. Hecht is saying that there is context, but if 95% of deepfakes on the internet are non-consensual pornographic images of women, how is the AI not recognising dangerous content immediately before it is uploaded? I still do not think we have an answer to that question.

There is not always a straightforward assessment of content. When it comes to deepfake and pornographic content, we absolutely prohibit and do not allow pornographic content, as well as non-consensual intimate imagery.

What does Mr. Hecht mean by "do not allow"? Is it in TikTok's guidelines or does it take accounts offline?

It is in our guidelines, and we are also increasingly improving our systems to make sure we address that content.

If a video is depicting gender-based violence or illegal activity on our roads or anywhere else, and TikTok has identified that and managed to delete the video within 24 hours or whatever, does TikTok then delete the account? Is that account then banned from TikTok forever?

It would depend on the specific violation. When it comes to-----

Why would it depend on anything? If illegal content is being shared, reshared and pressed and the algorithm is recommending all of this stuff, which is happening according to the DCU report I referenced earlier, why would those accounts not be deleted permanently?

When it comes to egregious harms and behaviours like non-consensual intimate imagery, we do immediately ban accounts that post that content. We have zero tolerance for that behaviour, and we outline in our guidelines what those areas are.

Will TikTok ban somebody with that email address from setting up another account under a different name? Is Mr. Hecht confident that person who posted that will never be able to be on his platform again?

We seek to prevent users who have engaged in egregious behaviours from ever coming back to our platform.

AI is very intelligent. There should not ever be a reason either that any of this content gets online in the first place or for that person to post again.

I want to move onto Coimisiún na Meán. It would be great if Mr. Godfrey could clarify something the Chair mentioned around publishing and the difference between these images and videos appearing on TikTok and if they were to appear, for example, on an RTÉ Instagram or TikTok page or a commercial station like Newstalk's page. What is the difference? There is no way that content would ever appear. It definitely would not appear more than once. It would not appear 42,000 times or 16,000 times. What is the difference between the two in terms of the regulatory process?

The legal framework for online social media companies contains a liability exemption for them for user-posted content. It is a conditional exemption. They are not liable for user-posted content if they do not have actual knowledge of it. Once they are made aware of it, they lose their protection from liability unless they take action to remove it as quickly as possible. That is the legal position.

What is the greatest hamstring for Coimisiún na Meán in regulating this type of online behaviour?

Coimisiún na Meán is trying to regulate yet it only has six cases open. What is Coimisiún na Meán most hamstrung by when trying to control this content, which is very damaging and harmful?

We operate within the legal framework and the resources we have.

What is Coimisiún na Meán most hamstrung by? Is it the legislation?

We have a very litigious environment, so we have to move carefully. Every investigation we have opened has had an immediate judicial review of the investigation-----

It is a dangerous place to be, to be afraid of litigation instead of protecting citizens online.

We are not afraid of litigation.

That is what Mr. Godfrey said Coimisiún na Meán was most hamstrung by. That was his answer.

We operate within the legal framework. We evaluate. If we have concerns and have reason to believe there is non-compliance, we take action. That is not a quick overnight process because of the legal protections everybody has. That is a matter for lawmakers really.

I only have a few seconds left. It is very clear from this meeting that something is not working. Videos can be uploaded pretty freely. If they are caught, they might be taken down or might not and accounts may or may not be banned forever. There is a regulatory and legal framework which Coimisiún na Meán is hamstrung by. Has the Department begun to reassess this whole area in the past couple of months and what it needs to look at in terms of regulation or legislation?

We always keep an eye out to see what more needs to be done. We told Coimisiún na Meán if it wants additional legislative measures we are happy to look at them. The Senator might be aware that in the Broadcasting (Amendment) Bill 2026, there are additional powers for Coimisiún na Meán. We are open to looking at other measures if required but they have to fit into the European legal framework. It is not just about additional measures, in that where there is regulation at European level, it does not allow national initiatives covering the same territory. That is the legal framework-----

It is something we have to jump out of.

-----we have to operate in. We are looking at it. We can see whether something needs to be added in legislation, secondary legislation or by Coimisiún na Meán in its online safety code but we do have some constraints.

That is an important point. I thank Ms Quill for saying it. It is a consideration for all of us. We need to jump out of that as quickly as possible because we cannot be hamstrung like that. We are not protecting our people.

I thank the Cathaoirleach for backing up my request for the viewing figures which will probably be one of the key takeaways from today. We need to know how many people saw this harmful illegal activity. I would like to see the community guidelines. It seems nothing can possibly break these community guidelines - not somebody being murdered and not somebody going the wrong down a motorway potentially killing themselves or others. It is very soft language - "non-compliance" instead of "breaking the law", "most appropriate" action instead of "really serious penalties". These are people's lives we are talking about - children's lives. There is psychological damage being done not just to our children. What is seen can never be unseen and it is not being caught fast enough. There are far too many grey areas. Coimisiún na Meán said that depicting a crime is not illegal. I would like Coimisiún na Meán to elaborate on that. TikTok mentioned that a wrong-way driving video can somehow be seen as educational. I would like both of those statements clarified or elaborated upon.

In our opening statement, we said that content depicting a crime is not in itself necessarily illegal. For example, if there is a public order incident, there may be footage of that public order incident carried on the news and posted online as a news report. It may be that people discuss the public order incident. There can be content that depicts a crime but does not encourage or glorify crime. It is important to draw a distinction between content that may depict a crime but is a legitimate exercise of freedom of expression and content that is used for purposes like encouraging or glorifying crime.

I think what we are talking about here is probably context. Context across all of these platforms is key. I was a journalist for 30 years - I know about quoting, fair comment, etc. but if you show something with the idea that, "Oh, look what this is" and "Oh, you should not do that" we still have to try to knuckle down on context and what is shown. I think I will get the same type of answer but I will allow Mr. Hecht to answer.

We care about context. It is incredibly important when somebody is depicting a behaviour that violates our community guidelines. We err on the side of caution. If it is not clear that an individual is uploading that footage in order to condemn it, to share news or report on that, we err on the side of caution and consider that to be promotional. When somebody not in an evasive way makes clear they are genuinely raising awareness and reporting on footage, we want to ensure that free expression is able to happen on our platform. We feel it is important for communities. We do err on the side of caution but we want to make sure that we hold on to that right.

It is far too soft. Erring on the side of caution is just simply not enough. Things should be removed until deemed safe, not left up until deemed unsafe. That is not erring on the side of caution. Remove it until due process is done - contact the person and maybe find out the context but remove it from view until then. Could that not be thought about?

When our systems or moderation process assesses a piece of content, we make a determination about whether it is promotional or whether it is done to report on and share news about an event. It is in that determination that we decide if that content violates our community guidelines.

Is that done by AI or by people?

It is done by our overarching moderation ecosystem which includes AI as well as humans.

I do not know what an "overarching ecosystem" means. Say it in English, please.

It includes automated moderation as well as human moderation. It depends on the specific context.

Is it still left up until that is looked at?

All content is moderated at the point of upload.

Will Mr. Hecht answer the question? Is it still left up so that I, my daughter or my son could switch it on - children could switch it on? It is still left there until TikTok decides it is illegal or whether it should be viewed?

Once we moderate a piece of content and identify it as violative of our community guidelines, it should be removed from the platform.

It is clearly not in many cases because we have all seen so many other videos. We could not even go into some of the videos we have seen.

I want to quickly go back to what happens then. What is the due process? If something is deemed illegal, what happens? Who takes it to the Garda? Who takes it to the person who has put it up and makes sure they will never see the light of day again on social media?

On the previous point, where a piece of content is depicted that we consider to be an imminent threat of harm or an imminent threat to life, for example, in the case of suicide, that immediately goes to An Garda Síochána. We have to make a determination around the type of content.

How long would that process take?

It could be minutes; it could be hours. We try to do it within two hours.

Did TikTok manage to get any of the data back it had requested in terms of the numbers?

The team is still working on it. We are trying.

TikTok may have this because it is probably a pop-out figure. What was the viewership of the most viewed dangerous driving video before its removal? TikTok surely most know the most viewed. I know it does not have the combined number.

That is not something that I have at hand but we have the team working to look into the data. We will add this to what they are looking at.

Has TikTok ever recommended dangerous driving content to someone who did not search for it? The reason I ask is that I have never searched for dangerous driving content yet this stuff has appeared on my timeline and my algorithm.

I cannot speak specifically to whether that happened but our system and our efforts are designed to remove that content from the platform.

If we have not identified content as violative, there is a possibility that it would be recommended but we also have additional checks and balances in place that as more views might be taking place on piece of content, it undergoes additional rounds of moderation. Our goal is to ensure that violative content is not seen by any users and that is where our efforts are driven.

When dangerous content is put up, the more views it gets, the more financially beneficial it is for TikTok. That is just the reality of the platform. Am I correct in saying that?

I spoke to this bit earlier. Our goal is to make TikTok a safe and inclusive platform. We want users to feel comfortable-----

I hear that but it wants to make returns as well.

We consider upholding safety for users as well as communities as a key part of the way we make TikTok a platform that people want to return to and engage on. The trust and safety department is at the centre of our work at TikTok to make the platform trusted and secure.

It was mentioned earlier that $2 billion is spent on online safety by TikTok. What amount is spent in Ireland specifically out of that $2 billion? There are 2.4 million customers.

On that specific number, our large investment in our Irish office is in our team and three quarters of our team are in the safety team. That is where the largest investment from TikTok Ireland goes.

They are based in Ireland but they are doing the safety worldwide.

Potentially, but it is a global service. They are based in Ireland, and they are paid in Ireland.

Does Ms Moss have a specific breakdown in terms of the Irish figures?

Regarding investment in Ireland, it is a global service. That is why we refer to global numbers.

If I had a shop here and a shop there, I know which shop is the most profitable.

The investment that we make is a global investment. It is impactful to all our community, not just our community in Ireland.

Does Ms Moss know what the figure is from a European perspective to broaden it out?

Again, it is a global service, so the investment we are making is for a global service that affects all our users equally.

That is after blowing my mind.

The RSA has come out this morning and said that all social media companies should stop the videos going up in the first place rather than acting after the fact or after the horse has bolted. That is a very fair recommendation. It is fair given the technologically advanced world that we live in. TikTok made a lot of references to the 24-hour figure in the opening statement. The witnesses said that 89% of content was removed within 24 hours. Why is 24 hours the benchmark for content when the witnesses accept, and we all accept, that it could potentially encourage behaviour that is going to put people's lives at risk? Why is it 24 hours?

We have several areas that we benchmark ourselves against. We care about 24 hours but we also care about whether we remove the content before anybody has viewed it. The data that we have shared is that around 83.4% of the content in the category that I have been speaking about around dangerous activity and challenges is removed before anybody views it. Our goal is to remove content, including content that violates our community guidelines, before anybody sees it. We also reference and benchmark against how much we remove content before it is reported because we want to make sure that we are being proactive and that we are removing that content before it can be seen on the platform.