Communications and Digital Committee

Uncorrected oral evidence: The Online Safety Act: implementation and impact

Tuesday 10 November 2026

2.30 pm

 

Watch the meeting

Members present: Baroness Keeley (The Chair); Viscount Colville of Culross; Baroness Elliott of Whitburn Bay; Baroness Fleet; Lord Holmes of Richmond; Lord Kirkhope of Harrogate; Lord Knight of Weymouth; Lord McNally; Baroness Owen of Alderley Edge; Lord Storey; Lord Tarassenko; The Lord Bishop of Winchester.

Evidence Session No. 6 Heard in Public Questions 111 - 133

 

Witnesses

Professor Clare McGlynn, Professor of Law, Durham Law School, Durham University; Sophie Mortimer, Head of Services and Revenge Porn Helpline Manager, SWGfL; Janaya Walker, Interim Director, End Violence Against Women Coalition; Alexandra Whiston-Dew, Partner, Mishcon de Reya LLP.

 

 USE OF THE TRANSCRIPT

  1. This is an uncorrected transcript of evidence taken in public and webcast on www.parliamentlive.tv.
  2. Any public use of, or reference to, the contents should make clear that neither Members nor witnesses have had the opportunity to correct the record. If in doubt as to the propriety of using the transcript, please contact the Clerk of the Committee.
  3. Members and witnesses are asked to send corrections to the Clerk of the Committee within 14 days of receipt.

Examination of witnesses

Professor Clare McGlynn, Sophie Mortimer, Janaya Walker and Alexandra Whiston-Dew.

 

  1.        The Chair: Good afternoon and welcome to this meeting of the Communications and Digital Committee. We are holding two evidence sessions today as part of our inquiry into the Online Safety Act. I am very pleased to welcome our first group of witnesses, who will be sharing their insights on how the new measures introduced under the Act have impacted the experiences of women and girls online. Before we move to questions, I ask each of you to introduce yourself and your organisation.

Alexandra Whiston-Dew: I am a partner at Mishcon de Reya.

Janaya Walker: I am the interim director of the End Violence Against Women Coalition.

Sophie Mortimer: I am head of support services at the charity SWGfL.

Professor Clare McGlynn: I am a professor of law at Durham University.

  1.        The Chair: This is the first of two panels today and, with four witnesses, we do need concise answers, so I hope you will help me with that. Please do not feel you have to respond extensively to any question where another witness has already responded in a way that you would support. Obviously, if there are differences, let us draw those out.

Our question areas will be around the impact of harmful content on women and girls, your assessment of Ofcom’s guidance, and whether the Online Safety Act and Ofcom’s codes address the design features and business models that facilitate violence against women and girls. There is a question on Ofcom’s enforcement, the pace of the new harms affecting women and girls, and improving the effectiveness of protections for women and girls.

I will ask the first question. According to a recent statutory report from Ofcom, which I am sure everyone has seen, there has been no significant change in children’s exposure to harmful content nine months after the children’s safety duties under the Online Safety Act came into force. So would you say that the picture is similar when it comes to online harms that disproportionately impact women and girls?

Alexandra Whiston-Dew: At Mishcon de Reya, we have seen cases that come through our doors that deal with very serious harm to women and girls online. We have not seen a change in terms of a de-escalation of those problems. We suggest that the issue is endemic and very difficult to address, particularly when there are online platforms that have no intention of complying with the legislation. We think that is serious and important to address.

Janaya Walker: We at the End Violence Against Women Coalition would agree that it is a similar picture, if not potentially worse. It is worth saying that we support the need for legislation and regulation in terms of the basis of the Online Safety Act. But the concern is that, in the years that have passed, there has been a real problem in both the implementation of the Act itself and its limitations, as well as, in parallel with that, the rapid expansion of emerging forms of harm, such as AI-generated harm.

We have carried out recent research with Refuge in a report called Dangerous by Design, which found that one in three women reported experiencing online violence against women and girls. Alarmingly, that rose to almost half for women and girls aged 16 to 24 in particular. To give another specific example, Refuge reports that it has seen a sustained increase in referrals through online platforms, including social media, since the Online Safety Act came into force. In the years ending March 2024 and 2025, it had an average of around three referrals per week involving social media. In 2026-27, that has jumped already to five to six a week. That is in the timespan since the Act came into force. We also refer to the research by the new national centre for VAWG and policing, which similarly found a rise in tech-facilitated gender-based violence.

It is also worth saying very briefly, in relation to the origin of the question on the picture for young people, that the exposure to harm for all children is unacceptable, but this is also gendered in and of itself. So rates of child sexual abuse are disproportionately towards girls, particularly AI-generated abuse and the ways that algorithms are feeding very gender-regressive views to girls and boys, underpinning and upholding some of the social norms and attitudes that bolster violence against women and girls in our everyday lives.

Sophie Mortimer: Within our particular area of work, it has started to but we cannot yet demonstrate it. The Act made intimate image abuse a priority offence, and Ofcom published its guidance last November. Since 30 September this year, the codes have recommended hash matching. Those are really important foundations, and we acknowledge that. But anecdotally we can say that, since the guidance came in, platforms have not made meaningful change with regards to uplifts to intimate image abuse mitigation measures.

There has been an uptick in integrating hashing via StopNCII since the code was issued, and that perhaps demonstrates the effectiveness of mandated codes over optional recommendations. But the harm is still growing exponentially. In 2025, the helpline reported 20,841 images for removal, which was up 34% on the year before, and demand has continued to rise every year since 2019. Between November 2025 and October 2026, we have reported 19,300 images, and 95.7% of those were of women. The awareness and capacity are growing, but this harm is running wild.

It is crucial to remember, of course, that this harm is significantly underreported both to police and to helplines. Modelling that we did on 2023 data indicated that approximately 369,000 UK women would experience intimate image abuse every year. For every woman we helped, there were 53 we did not. With all the numbers rising, we have to say that the picture in 2026 can only be worse.

Professor Clare McGlynn: As with my colleagues and many others who have given evidence before the committee, I unfortunately think that the potential of the Online Safety Act to bring about transformative change has not yet been realised. I will briefly mention one area that has gone under the radar: the role particularly of search engines in facilitating online abuse and harms. We know this from deepfake abuse particularly, and from Google’s role in facilitating deepfake abuse. We have seen some decisive steps in that area, thanks of course to the work of Baroness Owen, campaigners like Jodie and others.

On pornography, for example, we still see the search engines facilitating returns of illegal content. For example, if you search for rape pornography, it is downranked but it is on the 10th page. Force porn is downranked but it is on the third page, and Google does not downrank unwilling porn. If you type in “unwilling porn”, it returns all the porn sites. It downranks “sleep” content but not “asleep” content. If you search for incest porn, it returns all the porn sites. So the search engines are supposed to minimise this content that is normalising violence against women and girls, but it is a wholly inconsistent approach. They are doing just enough to make it look like they are taking action, but they are still enabling this violence and abuse to continue.

  1.        Lord Storey: What is your assessment of Ofcom’s guidance on protecting women and girls online? To what extent are platforms implementing its recommendations? I guess I know what you are going to say.

Professor Clare McGlynn: The guidance against women and girls is very welcome because it has some good measures of best practice in it, but of course it is not legally binding. Even though it is not legally binding, it was such a struggle to even get that in the Act, as many of you will be aware. There are steps that could be taken. We could make the guidance into a code of practice, but you are all very aware that even a code of practice is not necessarily a panacea because of the way those codes have been interpreted. So another option would be to include VAWG as one element of a risk assessment that platforms have to carry out. We are on track, but we really need to make these provisions legally binding.

Sophie Mortimer: We are seeing some progress but not enough, related to StopNCII.org, our platform. When the guidance was published, there were 17 industry partners, but now there are 30. Participation is very low. Ofcom estimates that about 10,000 platforms fall within scope. So we are currently at less than 1% of platforms that have signed up, and this is of course now mandatory since September. In 2025, more than a quarter of images were located on adult sites and around a fifth were in private channels such as email and messaging, most of which hash matching does not reach. Some platforms have told us that they do not think that the new measure applies to them because they do not consider themselves to be high risk through their risk assessments. We have voiced concern to Ofcom and industry that platforms are not accurately capturing their risk level for NCII on their services. Effectively, they are of course marking their own homework because they make these assessments themselves, and services that have the most need to act have the least incentive to conclude that they are high risk.

Janaya Walker: I largely echo what Clare has already said in terms of the fact that the guidance is relatively good; it is fairly ambitious and quite comprehensive. There was some good consultation with VAWG organisations broadly, but the major challenge that we are up against is that it is not binding in the same way as the code of practice, and this is a real challenge for us. When we think back, the basis of so much of the parliamentary debate around the Act was this widespread consensus that these platforms are not going to regulate themselves and are not going to do things voluntarily. But when it comes to violence against women and girls, so much of the good practice is set out in a voluntary piece of guidance. So that is definitely something that we would advocate to be upgraded, and with some urgency.

Alexandra Whiston-Dew: The only other point I would add is a spotlight on anonymous image boards. The point is well made in terms of search engines but, in terms of how the guidance operates, if it is not mandatory, that is one major issue. But the second point is the resistance against any kind of self-regulation. Where you have platforms that are designed to host illegal content and that is their sole purpose, they of course are not going to engage in any kind of self-regulation. There is a real lacuna there as to what can be done with those types of platforms. That is the spotlight that we would encourage you to look at.

  1.        Lord Storey: When Ofcom came to us, it said that the guidance was put together in a very collaborative way, which is good news, but I guess all four of you would suggest that it becomes statutory. Would that be fair to say?

Janaya Walker: Yes.

  1.        Lord Storey: You have already said why. I was just intrigued, Clare, in your comment that they just do enough to be under the radar. Could you amplify that a little bit more for me?

Professor Clare McGlynn: With the search engines, it was during the passage of the Online Safety Bill that Google finally downranked rape porn because before then you were able to search for that. So they downranked it to page 10. But it is a wholly inconsistent pattern because, for example, force porn you can find on the third page and rape porn is downranked to the 10th. They block unwanted porn but not unwilling porn. If someone spent a little more time thinking, “I really want to minimise the risk of this material appearing”, they would take a systematic approach. They would downrank unwilling and unwanted, and downrank “asleep porn” as well as “sleep porn”, because it just means that it is easily findable. So some of them are downranked but just not systematically, wholly or consistently.

  1.        Lord Storey: Do you think Ofcom was mindful of the fact that this was a huge area of work to do and, for whatever reason, decided, “Let’s go with guidance”? There seems to be a coalescence around that, but the next stage would be to make it statutory. Would that be a fair assessment?

Professor Clare McGlynn: During the legislative process, the campaign that many of us here were involved with was to make it a code of practice. But at the time, the Government only agreed to making it guidance. That was better than nothing because until then there had been nothing in the Bill about women and girls. But, yes, we are now in the situation of having guidance that is not legally enforceable. As you all know, in relation to Ofcom, everything takes so long. We are still waiting for these legally binding provisions.

  1.        Lord Storey: Your figures are devastating, are they not?

Sophie Mortimer: Absolutely. We passed our number of reports for the whole of last year before we reached the end of September. To follow Clare’s point, hash matching is mandatory to thousands of platforms. Yet platforms are still not doing that. So mandatory has to be the basis. Guidance clearly is not going to cut it.

The Chair: Thank you. I will have to move us on.

  1.        The Lord Bishop of Winchester: Do, in your opinion, the Online Safety Act and Ofcom codes adequately address the design features and business models that facilitate violence against women and girls, and what specific changes to platform design will make the greatest difference? Could Ofcom require them under its existing powers?

Janaya Walker: We would say no, it does not go far enough with regards to the design features and the business model. Connecting it to the earlier point, what we were up against when the Bill was going through was that we were told that violence against women and girls would be addressed in the Bill because you have a list of priority illegal content offences, some of which are forms of violence against women and girls, like stalking and harassment, for example. But our contention, and it remains the same, is that the way in which that sets up the regulator and platforms to assess what is happening is to look at particular pieces of content and think about how you mitigate them, but without thinking about the continuum of violence against women and girls, and the broader context of abuse. It is going to be a bit of a mixture in terms of the grey area of what reaches a criminal threshold. We have been advocating for a safety-by-design approach. We support the proposal put forward in terms of a safety-by-design code of practice, which would take the legislation further in terms of clearly articulating what safety by design means and how you can ensure that platforms are required to carry out measures, particularly preventative measures.

We found that Ofcom’s interpretation of the Act has a limited focus, and a lot of it is about measures after the harm has already occurred, rather than thinking about abuse and how it can be mitigated from the outset. It is worth reflecting on the impact of the VAWG guidance overall. Just a month after the VAWG guidance was published, we saw X’s AI tool Grok creating thousands—millions—of deepfake images, and 40% of them were sexualised images of women. That speaks to the fact that clearly there is not enough of a precedent set for foreseeing what the risks might be and how the harms in particular might disproportionately impact women and girls, and other marginalised groups.

  1.        The Lord Bishop of Winchester: I am asking you to speculate here. I was struck, Clare, by what you were saying about the linguistic distinctions between those categories. It seems to be quite a leap to suggest that Google is incapable of making those distinctions. It seems that there is an exercise of choice going on here. Would you agree? You are nodding.

Professor Clare McGlynn: Yes, absolutely.

The Lord Bishop of Winchester: They can design but they are not doing so. Is that fair?

Professor Clare McGlynn: Yes, I think so. If you were trying to think of the words that would be similar to, say, rape—rape porn is extreme porn and is not supposed to be on those platforms any more—you would think of words like “unwilling” and “unwanted”.

The Lord Bishop of Winchester: Or your algorithm would.

Professor Clare McGlynn: Exactly. We do not even need to get our thesaurus. It is just not happening. For example, although they now block “force” and “forced porn”, that only came after an intervention where I discovered you could search for “forced” and it returned results, but not “force”. So, although they blocked one, they had not blocked the other. It is a patchy, piecemeal response. Bing is just the same, by the way. Bing returns many of these searches as well. The fear, going into the future with further porn laws that we are supposed to be enforcing, is: are the platforms and search engines going to continue to facilitate this content?

Alexandra Whiston-Dew: I just wanted to add a point that the search engines are such an important party because, from a victim perspective, that has a significant impact. That is obviously clear to everyone. But if we are considering the restricted resources of Ofcom, for example, then obviously those primary search engines are significant. But there are also different platforms in the chain of publication, which would also be an area of focus. So, when you are thinking about establishing a website and the publication of that kind of content, it is not just the search engines that are a big business in that generation of that content to the end user. There are different levels in the chain of publication that could also be a consideration.

When you are thinking about design specifications—I hear what my colleagues are saying here, which is incredibly important, especially if we are going to try and corral some type of interest and focus around consensus—I just want to spotlight again some of the outliers, or perhaps not so much outliers as we would like, which are those websites that are conceived and published for the purposes of publishing illegal content.

I just want to make an anecdotal reference to the labelling and the commentary that is published on those websites, which is actively derisory to the work of Ofcom and the Government in terms of trying to deal with this illegal content. They are spotlighting the fact that they are publishing illegal content, and this is going under the radar not only from the search engines but the other platforms in the businesses in the chain of publication. Ofcom needs to really pull all its resources together and attack that type of receptacle of that illegal content.

  1.        Lord Knight of Weymouth: Before I move on to enforcement, just to follow up where Mike Storey was going, if the guidance was made into a code of practice—was made statutory—given the session we are about to have around freedom of expression, I wondered whether, Clare, perhaps you could comment on whether that would be at a cost to freedom of expression if it were to be made statutory.

Professor Clare McGlynn: Freedom of expression is obviously about protecting all people’s right to freely express ourselves. What we find with online abuse in particular is that it impacts adversely on women’s freedom of expression. Living with the kind of ever-present threat of being deepfaked, for example, means that many women feel less able to comment online, for example. Research I recently did with Fiona Vera-Grey found that two-thirds of women restrict their activities because of the fear of victimisation. So it is not even once you have been victimised but the fear of victimisation, and that impacts our freedom of expression. So freedom of expression is absolutely something we need to protect, but it is not just about protecting the expression of the perpetrators; it is about protecting everyone else from being impacted adversely by their ability to express themselves online.

Of course, we have Article 8 rights to privacy that we need to protect, and particularly around online abuse, we are talking about Article 3 rights to protection from torture and inhuman and degrading treatment. We need to remind ourselves that these sorts of harms can be life-shattering, sometimes life-ending and extremely degrading. So it impacts there as well.

Alexandra Whiston-Dew: I am not sure that freedom of expression is really a balancing consideration when considering the level of content that we are looking at here. These are straightforward illegal offences, and freedom of expression is not usually considered in the balance of the question here. It is not that we are considering the publication of defamatory allegations, for example; this is a different focus at this point.

  1.        Lord Knight of Weymouth: Let me move on to what I am supposed to ask you about, which is your assessment of Ofcom’s enforcement of the Online Safety Act’s protections against violence against women and girls under its existing powers. Given this guidance, you could argue that there is not a lot, but I will let you make that case. Obviously, I am then interested in where we could do more, apart from perhaps what you have already said. I am particularly interested in the reporting and complaints procedures. Clearly, Sophie, you operate what many people use at the moment for that purpose, so I am particularly interested in your answer. Perhaps we can start with you.

Sophie Mortimer: It is quite early to say, certainly in relation to the hash-matching measure, which, as I say, came into force only in September. But I cannot stress enough the importance of robust enforcement to prevent significant harm to thousands of UK women every day.

We are concerned, particularly given Ofcom’s recent publication on the children’s safety duties, that we will see a similar lack of impact, as I think one of your questions referenced. The enforcement programme opened on 9 September and enforcement started on 30 September, and Ofcom has shown elsewhere that it can use its powers. It has fined a pornography site for age assurance failures. It is a bit early for us to say how that programme will work, but it requires it to engage robustly, and sometimes maybe not just issuing that fine—we may well need more. I think it has also acknowledged itself that business disruption measures could be more available to it and more quickly, because a lot of the measures that are available to it take huge amounts of time.

We would ask for more in speed and reach. Formal investigations take months but an intimate image spreads in hours. Again, it has already been referenced, but there are sites that do not comply. Services hosting this content are overseas, some have no reporting route and some are set up to host it. Around 6% of the images that we reported in 2025 are still available, and some sites will geoblock rather than take content down, but that just cuts us off from the ability to report that content.

  1.        Lord Knight of Weymouth: Sorry, Sophie; is this then a problem that Ofcom does not have the enforcement powers that it needs, or is it that it is not using what it has robustly enough?

Sophie Mortimer: It is a bit of both. I think it has some but it needs to hit hard and fast now that the enforcement is in place. Given that StopNCII has 30 partners when there are probably 10,000 in scope, that clearly shows a massive failing on the part of industry to engage with what it is supposed to engage with. On business disruption, as I say, Ofcom has said it itself. It is not quick enough—it is just so slow. So absolutely, more needs to be made available to it, but yes, it can also be better using what it has.

  1.        Lord Knight of Weymouth: On the complaints and reporting process, we have heard that to complain on Meta is nine clicks, which is not a user experience it would design if it was something important to it. We do not have an e-safety commissioner; perhaps we should—a witness last week suggested that. Others last week told us that we should have a civil redress system. Any views on that are very welcome.

Sophie Mortimer: Reporting routes absolutely should be better and clearer. It cannot ever be acceptable to put somebody through nine clicks, particularly when they are in all likelihood having to report to multiple sites, some of which will have no routes at all. There should be a minimum standard for how people can access that sort of support. I have lost my thread, apologies. Go to Janaya, please, if you do not mind, and I will come back.

Janaya Walker: To echo what Sophie said originally about that, we think it is both that Ofcom is not properly enforcing in the way that we would expect in terms of the powers that it has, and that more regulation is needed in terms of addressing online violence.

  1.        Lord Knight of Weymouth: Is there anything specific? Specificity is really helpful to us in terms of what we might want to recommend.

Janaya Walker: We have quite a list: upgrading the statutory guidance to a code of practice, introducing a safety-by-design code of practice, echoing more in the way of business disruption measures, and removing the safe harbour provision. As I think you heard from Lorna Woods, Online Safety Network has put forward a 10-point plan with some kind of technical amendments which we are also behind. I think there is a quite a consensus building in terms of the range of options.

Lord Knight of Weymouth: If there is anything on the list that you then want to follow up by sending to us, that is also possible.

Janaya Walker: Absolutely; we can do that.

The case I wanted to draw your attention to was in relation to your question of whether Ofcom is doing enough to enforce the powers that it has. In February of this year, we supported a survivor to initiate legal proceedings against Ofcom for its failure to take action against a platform which was hosting non-consensual intimate images, including child abuse images. In this instance, the survivor with the pseudonym Jane was in a process for almost a year and a half, trying to instigate action before we supported her, and I think it was within a matter of days after legal action was initiated with a pre-action protocol letter that Ofcom announced that it was now taking enforcement action against these sites.

The Chair: Can we move on from that? As I mentioned earlier, we do not want to talk about cases as they move forward.

Janaya Walker: Absolutely. That was just to demonstrate the lengths that survivors have to go to, and it also speaks to the second part of the question about the extent to which survivors are left with limited options for recourse. I think that goes to the question you asked about an e-safety commissioner. We are very supportive of an online abuse commission which builds on existing models where there is a head who can represent the interests of victims and survivors. We obviously have really crucial services such as the Revenge Porn Helpline and specialist women’s services, but they are so desperately underfunded. When you compare it with the scale of online violence against women and girls and the new forms of harms emerging, I think there is a real gap to ensure that there is somebody who can advocate for survivors, and that it is not so much of an onus on survivors to be across multiple platforms, as Sophie said, trying to address the harms that have been carried out. 

  1.        Lord Knight of Weymouth:​ Alexandra and Clare, do you want to add anything? 

Alexandra Whiston-Dew:​ One point I wanted to make is that I think we have to question what is the resistance to taking extremely decisive action. What is the resistance to allowing Ofcom to switch off content that is causing such incredible harm? I also want to illustrate, not by way of particular case study, although that is helpful to hear. Content that is published online usually is associated with commentary or potentially identifying content, which means that there is not only online harm in terms of reputational or mental health repercussions—or all the variety of issues that I am sure you have heard about before in terms of publishing intimate images online and pornographic images without consent—but also the physical harms that can materialise if someone is identifiable as the person in that image, and how that could proliferate in real life. I do not mean to say that the harm that is experienced by publication online is not in real life, but there is the physical harm that could be applied to that individual.

In those significant consequences, what is the resistance to allowing Ofcom, or some of the other factors that we have talked about, to be able to switch off that content, with appeal mechanisms following, analogous to ex parte injunctions in a civil proceeding? That is not available, as I understand it, to Ofcom right now and it is something potentially that should be considered, bearing in mind the real impact that people are suffering as a result. 

Professor Clare McGlynn:​ In relation to enforcement, there are two particular areas I would like to draw your attention to, and they relate to Ofcom’s failure to use its existing powers. The first is in relation to pornography. What we see, in essence, is that the transformative opportunity of the Online Safety Act has not been realised. I have talked about how rape pornography is facilitated through the search engines, but I just underscore that it is still freely accessible on the mainstream porn sites, including X, and Ofcom has not yet taken any action in relation to the content of extreme pornography, although it has obviously taken enforcement action about age assurance. What I would say there is that age assurance is very welcome but it does not deal with the content. You still have those who, through VPNs and suchlike, are accessing this extreme pornography, and the harm does not stop at age 18; the material is still there.

I think we really need to tackle that content. Rape pornography has not yet been acted on. Again, I have given the committee a few of the examples from X and Constance Kampfner from the Times has also done a recent investigation on that. My particular concern about that is that you will know that some of the pornography laws were updated last week to cover strangulation, incest and adults pretending to be children. Ofcom issued a statement to all the media last week saying that the obligations under the Online Safety Act for the services would not come into operation until it produced guidance on that. That is not what the Act says. That is my analysis of it and that of Lorna Woods from the Online Safety Network. Schedule 7 has been changed. Those are priority illegal offences. Those services need to now be acting in relation to that content, but Ofcom issued a statement saying the services do not have to do that until it issues its guidance, and that is at least 18 months hence. That is a really interesting example. I know you have heard examples in other sessions about where Ofcom’s interpretation of the Act is not similar to what is actually written in the Act, or what some of the rest of us can identify.

The other area of enforcement is in relation to Grok, which we have already mentioned. I was somewhat surprised, shall we say, when Ofcom itself used that as an example of its speedy response, because, yes, it launched the investigation on 12 January, but here we are and we still have no resolution to that investigation, nearly nine or 10 months later. My understanding is that on X with Grok you can still produce some of these intimate images without consent. We are just waiting now for the regulator to tell us that. So Ofcom is not yet using its powers in these areas in ways that I think it should, and I think there is evidence of it having a surprising interpretation of the Online Safety Act. 

  1.        Lord Knight of Weymouth:​ Finally, there is something particularly interesting there—everyone else has been interesting—which is around different legal interpretations. There is some concern, perhaps, that Ofcom’s lawyers may be interpreting differently to you, Lorna or whoever. How do we resolve that? How do we create the legal certainty so that they can get on with it? 

Professor Clare McGlynn:​ I understand where you are coming from in that. In relation to this, one example is that it just does not say anywhere in the Act that services have to wait for Ofcom’s codes. It is almost like it is not a difference of interpretation. I do not know where it is in the Act. I have asked it, but I have not yet found out. There are other areas—you are right—in which there may be more interpretation. I suppose what I would say, which, again, is what many others have said, is that the approach to interpreting it has been so minimalist, conservative with a small “c” and overly cautious, whereas many of us, including yourselves, when we had the Online Safety Act, were expecting a step change. We were expecting action against these platforms that were not regulating themselves so that we would no longer experience some of these harms, and that would have required a determination to implement the Act in the way it was intended. That is what we have not seen. 

The Chair:​ We have already had a shout-out for Baroness Owen, and quite rightly too.​ I come to her.

  1.        ​​Baroness Owen of Alderley Edge:​ Thank you. Importantly, I want to start by declaring my interest as receiving pro bono legal advice on intimate image abuse from Mishcon de Reya, which is important in this committee. I wanted to speak to Sophie specifically about the centralised hash registry, which we worked on at the beginning of the year, along with Charles, that came in under the Crime and Policing Act 2026. It gave the Secretary of State powers to make regulations and create a centralised hash registry. The Government are currently undertaking a scoping exercise on this. Do you know where it is up to? Please could you update the committee? 

Sophie Mortimer:​ The scoping exercise has just reached its conclusion, I understand. I sit on the working group with the Home Office; it is drawing together conclusions and then recommendations will be made to Ministers for next steps, so we will be encouraging those next steps to be taken. 

  1.        Baroness Owen of Alderley Edge:​ Also in the Crime and Policing Act, we introduced a 48-hour takedown, with the help of Professor McGlynn and your help, Sophie. Along with the 48-hour takedown and the hashing measures which Ofcom is introducing, where will the gaps remain? How much will this help victims, and where do you see the remaining gaps? 

Sophie Mortimer:​ We have definitely made some progress, and thanks to you, Baroness Owen, and others for your support with that. Hash matching works. It stops known images being uploaded again on participating services. It is privacy-preserving, because the image never leaves the victim’s device, and it moves the burden from the victim towards the platform. The 48-hour duty is in the Crime and Policing Act, and Ofcom has said it will consult by the end of the year on the codes, so victims still cannot rely on that in practice.

The gaps as we see them are these: first, scope and self-assessment. The measure is tied partly to size, so some high-risk services fall outside it, and it relies on services’ own risk assessments. Some platforms have told us that they do not think it applies to them because they do not consider themselves to be high risk so, again, they are marking their own homework. Private messaging, of course, is largely out of reach.

The second is equivalence. Services using their own in-house systems might not receive hashes from victims created through StopNCII or verified hashes from other platforms’ own bespoke services. The third is existing copies. Hash matching blocks new uploads. It does not find what is already online, and we would ask that each service checks existing content against every new hash, starting with exact matching, so a victim only has to report once.

The fourth is hidden harms. Victims of voyeurism and synthetic images often do not know that images exist. They have no original to hash from. So verified hashes from platforms and trusted organisations really do matter.

Fifthly, there are sites that do not comply. As already referenced, around 6% of content that we report remains online, usually overseas or on purpose-built services that simply ignore our reports. Removal duties cannot reach them, and that is where it gets really important that Ofcom uses those business disruption measures that it has and that it is given more robust and faster ones.

The 48 hours should be a backstop, not a target. Removing a reported image is quick but finding altered copies is slower. A single short deadline for both pushes services to exact matching and leaves altered copies unaddressed. We suggest treating reported content and near duplicates separately, with a fixed outer limit that is never open-ended, and shortening deadlines in stages as services show that they can meet them

Faster removal needs safeguards against mistaken or malicious reports. No type of online content is removed instantly today, not even child sexual abuse material. The fastest reliable route is reports from trusted organisations that have verified the content, which depends on those organisations having the capacity to meet the demand.

  1.        Baroness Owen of Alderley Edge: Professor McGlynn, do you have anything to add?

Professor Clare McGlynn: No, except to talk about the rights of an individual. This is something that I raised in one of the committees that your Chair was on during the passage of the Bill: an individual still does not have the right to demand an image be taken down. That is where we just lack the capacity to say to an individual, “This is what you need to do. You can bring this legal action against X, for example, the perpetrator, and get the material deleted”. British Columbia in Canada has a swift online civil redress system where you can make these claims and get instant orders to get the material taken down, and you can even get compensation for some of the harms suffered. Other countries are doing more to give individuals that right, as well as all the other measures at platform level.

Janaya Walker: I absolutely echo what Clare said about the recommendation for a commission and building on existing models elsewhere, particularly given the points that I raised around the options that are available for survivors.

Sophie mentioned the “trusted flagger” model. That is really important, as she said, but I emphasise how much that relies on the capacity and resources for those specialist organisations. EVAW is a member organisation, and so many of our members reflect the fact that they are not funded for these forms of support when it comes to the complex case loads that are produced when you are thinking about online violence against women and girls.

It is worth remarking that Refuge’s “trusted flagger” relationships with services have been very successful, but there are suggestions that, since the guidance came into force, they have been deteriorating in terms of the response of platforms. It has said that 53.7% of content removal requests used to be successful but that has now dropped to less than 20%, to 18.2%. I will look at specific services: since November 2025, no content removal requests made to Facebook via Refuge’s “trusted flagger” service have been successful, and it is at only 40% for content removal requests to TikTok. That is a real concern when we are talking about the behaviour of the platforms themselves and their adherence to existing regulation, and about what that means for the specialist services and their capacity to support survivors in this limbo and, without the likes of an online commission, to advocate for victims’ and survivors’ interests.

  1.        Baroness Owen of Alderley Edge: Has Refuge speculated as to why that is?

Janaya Walker: It has not. The challenge has been about the relationships with the platforms themselves. We would look at the broader context of the research showing that trust and safety provisions on some platforms have reduced, and some of the online safety protections have gone down recently in the context that we are in.

The Chair: It is worth mentioning that we discussed civil redress last week. You have probably heard that Damian Collins went back to the Joint Committee, which recommended a clearer route for civil redress through the courts for victims, supporting what you have said. Baroness Kidron, who was also a witness, made a case for an independent e-safety commissioner, and we know that the Law Commission is looking at redress as part of its extending product liability law. So those points are well made.

  1.        Lord Kirkhope of Harrogate: I am going to ask you about harms. You will remember that the 2019 White Paper on the subject specifically said that the list of harms would be “neither exhaustive nor fixed”, looking ahead. I know that Sophie, in commenting, said that one of the worst harms is a lack of action, as opposed to the specifics. Looking at the regime as a whole, the developments in technology and all the other movements that are going on, to what extent has the regime kept pace with the new harms? Looking also at the priority areas here—although you have talked about this a bit already—can each of you give me your views on whether you think we are keeping up with the new harms, what those new harms are and which ones you think should be given the priority?

Professor Clare McGlynn: In principle, the Online Safety Act ought to have been able to keep up with new and emerging harms because the point about dealing with the systems and processes is that those platforms will identify emerging risks and take steps to mitigate them. That relies on those platforms properly investigating the risks that are there, and that is why we have seen some of that fall down.

One of the most obvious areas is in relation to AI chatbots. There is a failure to properly deal with that issue under the existing measures in the legislation, but there are also failures because the legislation does not cover all the instances of chatbot harm. For example, my colleagues and I did some work on how chatbots are reshaping violence against women and girls, and there is one area that we call chatbot-simulated abuse, where you design the character that you will engage with. You can get on platforms like Chub AI, which has 10 million monthly users, and, when you want to create a character, two of the options that drop-down gives you are incest and rape. You can create that character and engage in that conversation, and that is not yet illegal content. Even the Government’s change to include chatbots in the scope of the Online Safety Act would not cover that sort of harm. We need to address that in proper regulation for AI and chatbots.

Sophie Mortimer: Thus far the law has followed each new tool, rather than setting standing expectations. We have had an offence for creating deepfakes and another for supplying nudification tools after episodes that really caught the public’s attention. These are welcome but the approach is always a step behind. There needs to be more engagement with front-line services that are seeing these patterns early. We have reported 96 cases involving synthetic sexual content in the 12 months to October 2026, compared with 63 in the whole of 2025. Given that victims often do not know that this content of them has been created, these are really just the tip of the iceberg. Callers also are increasingly telling us that their first source of advice is a generative AI chatbot, which brings people to specialist help, but that raises concerns about the accuracy of the advice they have been given.

Janaya Walker: Similarly to what Clare said, I would reflect that it comes down to the way in which Ofcom has interpreted the Act’s safety-by-design provisions. More clarity is needed there. Otherwise, as Sophie set out, we are in a situation where we wait until the harm has occurred, there is public outrage and survivors are harmed, and then action is taken to address that particular form of content, rather than thinking about what we can do within the legislation and regulation to foresee these kinds of risks.

It is difficult because on a month-to-month and yearly basis we are seeing the emergence of new forms of harms. We have talked about AI chatbots and AI-generated abuse. We have also been dealing a lot this year with the fallout of so-called wearable technologies, with the likes of Meta’s spy glasses—or smart glasses, sorry. There is a contention there about the extent to which these wearable technologies would fall within regulation. There is then a crossover, because women’s experience is that they are covertly recorded and then that is uploaded on to social media platforms, and again it is only after the harm occurs that action potentially falls under the scope of the Online Safety Act.

Ensuring that the safety-by-design code of practice was introduced would mean that we could get to a place where we did not have to constantly chase after the harm.

  1.        Lord Kirkhope of Harrogate: Deepfakes are a good example of something that has come on, as it were, in terms of the harm that has been identified and given priority. From a legal point of view, could you comment on the interpretation and how you deal with that?

Alexandra Whiston-Dew: The points that I was thinking may be relevant might be a bit controversial: there is a lot of legislation, some quite traditional, that applies to violence against women and girls, and the lag in the implementation of these specific offences can be problematic in the way that you have heard about. The merit of amending legislation, or devising and implementing new legislation, is public dialogue and understanding about what is happening to women and girls. That is valuable, but I question whether that is the most impactful way of doing it.

The second point is the police recognition and guidance that they adhere to when they are considering these new offences, and the conversation and training that they receive in order to address these new harms, which may be caught in the net of more traditional harms—a weird thing to say about criminal law, but I am talking about stuff that has been in situ for a long time. That helps people to understand what is happening to women and girls, and it starts a conversation and public dialogue that I think is helpful to survivors.

Making successive changes in legislation is difficult. Thinking about the level of specificity that is required in order to make it impactful, and making sure that it does not unintentionally cause issues, given the quality control process that is required, is a long-term project. Given the quick turnover of how things materialise, that is something to take into consideration.

  1.        Baroness Fleet: You have all been extremely helpful and constructive as well, so thank you. We have dealt with some of what I am going to ask you, but I would like each of you to choose two key recommendations that the committee should make to improve the effectiveness of our online safety regime to protect women and girls online. We have covered a lot of ground and some of it you have already discussed, but, if you had to choose two, which would be your priorities?

Professor Clare McGlynn: I will offer two that we have not yet discussed but which are a priority. One is in relation to pornography regulation. That would mean establishing an auditor or an additional regulator to make sure we can monitor the forms of extreme pornography that are available on these mainstream sites. That is something that Baroness Bertin recommended in her review and is under consideration, and it might make a considerable difference to the sorts of issues we have been discussing.

The second is in relation to app stores, which are not covered directly in the Online Safety Act regime. That came to mind when we were discussing deepfakes, because one of the ways in which deepfake abuse has so proliferated is through the easy accessibility of nudification tools in the app stores, which carries on. It is nearly 10 years since we first found out about the deepfake technology but we are still having to deal with the aftermath of that, so the app stores would be my second recommendation.

Sophie Mortimer: From a victim perspective, which I have to come from, my first recommendation would be a meaningful and effective one-and-done: the promise that services should act on a verified hash across their whole service once, including existing content, and hash matching should reach every high-risk service, whatever its size would be. My second would be action on non-compliant sites, clear timescales for Ofcom to act on “trusted flagger” referrals, a clear plan for business disruption and integration of the NCII register and a statutory basis for access providers, hosts and domain services to take action.

Janaya Walker: We have spoken extensively about the codes of practice. I would add the creation of an online commission to ensure that victims’ and survivors’ interests are better represented. I would add the ask that we have had for some time about ring-fencing a percentage of the digital services tax income for specialist violence against women and girls services. Both of those together are about reflecting the fact that women and girls have disproportionately borne the brunt of these failures by the platforms and the harms that they have produced.

Alexandra Whiston-Dew: I am searching for two extra beyond what has already been said. One is transparency in the process for Ofcom. We understand that some feedback is provided to complainants, but it is not extensive. People or complainants do not understand where in the process Ofcom is. Some transparency around that would be extremely helpful for everyone involved, not least because we can then spot problems where there may be some resistance in the system.

The second point to reiterate is related to the one I made earlier: what is the resistance to strong, effective, timely and impactful results? Why is Ofcom not addressing these issues to the full extent of the legislative capability that has been provided? What is the resistance to taking that step further, which is the switch-off capability with an appeal following switch-off?

The Chair: Thank you. You have been an excellent panel of witnesses. In particular, you have been concise with your answers even though there is an awful lot that you and your organisations probably want to say. You have been very helpful in that regard today.