ONLINE SAFETY BILL (Second sitting)

Damian Collins Excerpts

Committee stage (re-committed clauses and schedules)
Tuesday 13th December 2022 Public Bill Committees

[Paul Scully]

Hate crime legislation will always be considered by the Ministry of Justice, but I am not committing to any changes. That is beyond my reach, but the two shields that we talked about are underpinned by a safety net.

[Damian Collins (Folkestone and Hythe) (Con)]

Does my hon. Friend agree that the risk assessments that will be done on the priority illegal offences are very wide ranging, in addition to the risk assessments that will be done on meeting the terms of service? They will include racially and religiously motivated harassment, and putting people in fear of violence. A lot of the offences that have been discussed in the debate would already be covered by the adult safety risk assessments in the Bill.

[Paul Scully]

I absolutely agree. As I said in my opening remarks about the racial abuse picked up in relation to the Euro 2020 football championship, that would have been against the terms and conditions of all those platforms, but it still happened as the platforms were not enforcing those terms and conditions. Whether we put them on a list in the Bill or talk about them in the terms of the service, they need to be enforced, but the terms of service are there.

[Damian Collins]

On that point, does my hon. Friend also agree that the priority legal offences are important too? People were prosecuted for what they posted on Twitter and Instagram about the England footballers, so that shows that we understand what racially motivated offences are and that people are prosecuted for them. The Bill will require a minimum regulatory standard that meets that threshold and requires companies to act in cases such as that one, where we know what this content is, what people are posting and what is required. Not only will the companies have to act, but they will have to complete risk assessments to demonstrate how they will do that.

[Paul Scully]

Indeed. I absolutely agree with my hon. Friend and that is a good example of enforcement being used. People can be prosecuted if such abuse appears on social media, but a black footballer, who would otherwise have seen that racial abuse, can choose in the user enforcement to turn that off so that he does not see it. That does not mean that we cannot pursue a prosecution for racial abuse via a third-party complaint or via the platform.

--- Later in debate ---

As a result of the Bill, people will be able to stop seeing content on YouTube, for example, promoting eating disorders, but they might not be able to stop seeing content promoting online poker sites, when that might be causing a significant issue for their health, so not including that is bit of an oversight. As I say, user empowerment is important, but the Government have not implemented it in nearly as good a way as they should have done, and the Opposition amendments would make the Government amendments better.

[Damian Collins]

I rise briefly to say that the introduction of the shields is a significant additional safety measure in the Bill and shows that the Government have thought about how to improve certain safety features as the Bill has progressed.

In the previous version of the Bill, as we have discussed at length, there were the priority legal offences that companies had to proactively identify and mitigate, and there were the measures on transparency and accountability on the terms of service. However, if pieces of content fell below the threshold for the priority legal offences or were not covered, or if they were not addressed in the terms of service, the previous version of the Bill never required the companies to act in any particular way. Reports might be done by Ofcom raising concerns, but there was no requirement for further action to be taken if the content was not a breach of platform policies or the priority safety duties.

The additional measure before us says that there may be content where there is no legal basis for removal, but users nevertheless have the right to have that content blocked. Many platforms offer ad tools already—they are not perfect, but people can opt in to say that they do not want to see ads for particular types of content—but there was nothing for the types of content covered by the Online Safety Bill, where someone could say, “I want to make sure I protect myself from seeing this at all,” and then, for the more serious content, “I expect the platforms to take action to mitigate it.” So this measure is an important additional level of protection for adult users, which allows them to give themselves the certainty that they will not see certain types of content and puts an important, additional duty on the companies themselves.

Briefly, on the point about gambling, the hon. Member for Aberdeen North is quite right to say that someone can self-exclude from gambling at the betting shop, but the advertising code already requires that companies do not target people who have self-excluded with advertising messages. As the Government complete their online advertising review, which is a separate piece of work, it is important that that is effectively enforced on big platforms, such as Facebook and Google, to ensure that they do not allow companies to advertise to vulnerable users in breach of the code. However, that can be done outside the Bill.

[Kirsty Blackman]

My concern is not just about advertising content or stuff that is specifically considered as an advert. If someone put up a TikTok video about how to cheat an online poker system, that would not be classed as an advert and therefore would not be caught. People would still be able to see it, and could not opt out.

[Damian Collins]

I totally appreciate the point that the hon. Lady makes, which is a different one. For gambling, the inducement to act straightaway often comes in the form of advertising. It usually comes in the form of free bets and immediate inducements to act. People who have self-excluded should not be targeted in that way. We need to ensure that that is rigorously enforced on online platforms too.

[Kim Leadbeater]

It is a pleasure to serve under your chairship, Dame Angela. It is lovely to be back in a Public Bill Committee with many familiar faces—and a few new ones, including the Minister. However, after devoting many weeks earlier this year to the previous Committee, I must admit that it is with some frustration that we are back here with the Government intent on further weakening their Bill.

Throughout the passage of the Bill, I have raised a number of specific concerns, from democratic and journalistic exemptions, to age verification, recognised news publishers, advocacy bodies and media literacy. On clause 14, while I support the principles of Government amendments 15 and 16, I draw the Minister’s attention to the importance of amendment (a) to amendment 15 and amendment (a) to amendment 16. He has already said that he is sympathetic to those amendments. Let me try to convince him to turn that sympathy into action.

I will focus primarily on an issue that is extremely important to me and to many others: extremism and radicalisation. However, while I will focus on the dangers of extremism and radicalisation, be it right-wing, Islamist, incel or other, the dangers that I am about to set out—the chain of events that leads to considerable harm online—are the same for self-harm content, eating disorder content, health disinformation, climate change disinformation or any dangerous, hateful material directed at people based on their sex, sexual orientation, ethnicity, religion or other characteristics.

Such content is not just deeply offensive and often wholly inaccurate; it is dangerous and vile and serves only to spread harm, misinformation and conspiracy. To be clear, such content is not about a social media user stating how upset and angry they are about the football result, or somebody disagreeing legitimately and passionately about a political issue. It is not the normal, everyday social media content that most people see on their feeds.

This is content that is specifically, carefully and callously designed to sit just below the criminal threshold, yet that can still encourage violence, self-harm or worse. It is content used by extremists of all types that lures vulnerable people in, uses social media likes and comments to create the illusion of legitimacy and popularity, and then directly targets those most likely to be susceptible, encouraging them either to commit harm or to move on to smaller but high-harm platforms that may fall out of the scope of the Bill. This is not free speech; it is content that can act as a dangerous gateway to radicalisation and extremism. The Government know how dangerous it is because their own report from His Majesty’s Prison and Probation Service last year found:

“The Internet appears to be playing an increasingly prominent role in radicalisation processes of those convicted of extremist offences in England and Wales.”

Hon. Members will understand my deep and personal interest in this matter. Since the murder of my sister, a Member of this House, six and a half years ago by a far-right extremist, I have worked hard to bring communities and people together in the face of hatred. Some of that work has included meeting former extremists and discussing how they were radicalised. Those conversations were never easy, but what became very clear to me was that such people are not born extremists. Their radicalisation starts somewhere, and it is often somewhere that appears to be completely innocent, such as a Facebook group about issues or problems in their community, a Twitter discussion about current affairs or the state of the country, or even a page for supporters of their football team.

One day, a comment is posted that is not illegal and is not hate speech, but that references a conspiracy or a common trope. It is an ideological remark placed there to test the water. The conversation moves on and escalates. More disturbing or even violent comments start to be made. They might be accompanied by images or videos, leading those involved down a more sinister path. Nothing yet is illegal, but clearly—I hope we would all agree—it is unacceptable.

The number of contributors reduces, but a few remain. No warnings are presented, no flags are raised and it appears like normal social media content. However, the person reading it might be lonely or vulnerable, and now feels that they have found people to listen to them. They might be depressed or unhappy and looking to blame their situation on something or someone. They might feel that nobody understands them, but these people seem to.

The discussion is then taken to a more private place, to the smaller but more harmful platforms that may fall outside the scope of the Bill, but that will now become the go-to place for spreading extremism, misinformation and other harmful content. The radicalisation continues there—harder to track, harder to monitor and harder to stop. Let us remember, however, that all of that started with those legal but harmful comments being witnessed. They were clearly unacceptable, but mainstream social media give them legitimacy. The Online Safety Bill will do nothing to stop that.

Unfortunately, that chain of events occurs far too often. It is a story told many times, about how somebody vulnerable is lured in by those wishing to spread their hatred. It is hosted by major social media platforms. Hon. Members may remember the case of John, a teenager radicalised online and subsequently sentenced. His story was covered by The Guardian last year. John was feeling a sense of hopelessness, which left him susceptible to the messaging of the far right. Aged 15, he felt “written off”: he was in the bottom set at school, with zero exam expectations, and feeling that his life opportunities would be dismal. The far right, however, promised him a future. John became increasingly radicalised by an online barrage of far-right disinformation. He said:

“I was relying on the far right for a job. They were saying that when they got power they would be giving jobs to people like me”.

John now says:

“Now I know the posts were all fake, but the 15-year-old me didn’t bother to fact-check.”

For some people in the room, that might seem like a totally different world. Thankfully, for most of us, it is. However, if Members take the time to see some of that stuff online, it is extremely disturbing and alarming. It is a world that we do not understand, but we have to be aware that it exists. The truth, as we can see, is that such groups use popular online platforms to lure in young people and give them a sense of community. One white nationalist group actively targets younger recruits and recently started Call of Duty warcraft gaming tournaments for its supporters. Let us be clear: John was 15, but he could easily have been 18, 19 or indeed significantly older.

John was radicalised by the far right, but we know that similar methods are used by Islamist extremists. A 2020 report from New York University’s Centre for Global Affairs stated:

“The age of social media has allowed ISIS to connect with a large-scale global audience that it would not be able to reach without it...Through strategic targeting, ISIS selects those who are most vulnerable and susceptible to radicalization”.

That includes those who are

“searching for meaning or purpose in their life, feeling anger and…alienated from society”.

The ages that are most vulnerable are 15 to 25.

Social media platforms allow ISIS to present its propaganda as mainstream news at little to no cost. Preventing that harm and breaking those chains of radicalisation is, however, possible, and the Bill could go much further to put the responsibility not on the user, but on the platforms. I believe that those platforms need unique regulation, because social media interaction is fundamentally different from real-life social interaction.

Social media presents content to us as if it is the only voice and viewpoint. On social media, people are far more likely to say things that they never would in person. On social media, those views spread like wildfire in a way that they would not in real life. On social media, algorithms find such content and pump it towards us, in a way that can become overwhelming and that can provide validity and reassurance where doubt might otherwise set in.

Allowing that content to remain online without warnings, or allowing it to be visible to all users unless they go searching through their settings to turn it off—which is wholly unrealistic—is a dereliction of duty and a missed opportunity to clean up the platforms and break the chains of radicalisation. As I set out, the chain of events is not unique to one form of radicalisation or hateful content. The same online algorithms that present extremist content to users also promote negative body image, eating disorders, and self-harm and suicide content.

I hope the Committee realises why I am so impassioned about “legal but harmful” clauses, and why I am particularly upset that a few Conservative Members appear to believe that such content should remain unchecked online because of free speech, with full knowledge that it is exactly that content that serves as the gateway for people to self-harm and to be radicalised. That is not free speech.

--- Later in debate ---

[Paul Scully]

I am seeking to impose new duties on category 1 services to ensure that they are held accountable to their terms of service and to protect free speech. Under the status quo, companies get to decide what we do and do not see online. They can arbitrarily ban users or remove their content without offering any form of due process and with very few avenues for users to achieve effective redress. On the other hand, companies’ terms of service are often poorly enforced, if at all.

I have mentioned before the horrendous abuse suffered by footballers around the 2020 Euro final, despite most platforms’ terms and conditions clearly not allowing that sort of content. There are countless similar instances, for example, relating to antisemitic abuse—as we have heard—and other forms of hate speech, that fall below the criminal threshold.

This group of amendments relates to a series of new duties that will fundamentally reset the relationship between platforms and their users. The duties will prevent services from arbitrarily removing content or suspending users without offering users proper avenues to appeal. At the same time, they will stop companies making empty promises to their users about their terms of service. The duties will ensure that where companies say they will remove content or ban a user, they actually do.

Government new clause 3 is focused on protecting free speech. It would require providers of category 1 services to remove or restrict access to content, or ban or suspend users, only where this is consistent with their terms of service. Ofcom will oversee companies’ systems and processes for discharging those duties, rather than supervising individual decisions.

[Damian Collins]

I am grateful for what the Minister has said, and glad that Ofcom will have a role in seeing that companies do not remove content that is not in breach of terms of service where there is no legal requirement to do so. In other areas of the Bill where these duties exist, risk assessments are to be conducted and codes of practice are in place. Will there similarly be risk assessments and codes of practice to ensure that companies comply with their freedom of speech obligations?

[Paul Scully]

Absolutely. As I say, it is really important that people understand right at the beginning, through risk assessments, what they are signing up for and what they can expect. To come back to the point of whether someone is an adult or a child, it is really important that parents lean in when it comes to children’s protections; that is a very important tool in the armoury.

New clause 4 will require providers of category 1 services to ensure that what their terms of service say about their content moderation policies is clear and accessible. Those terms have to be easy for users to understand, and should have sufficient detail, so that users know what to expect, in relation to moderation actions. Providers of category 1 services must apply their terms of service consistently, and they must have in place systems and processes that enable them to enforce their terms of service consistently.

These duties will give users the ability to report any content or account that they suspect does not meet a platform’s terms of service. They will also give users the ability to make complaints about platforms’ moderation actions, and raise concerns if their content is removed in error. Providers will be required to take appropriate action in response to complaints. That could include removing content that they prohibit, or reinstating content removed in error. These duties ensure that providers are made aware of issues to do with their services and require them to take action to resolve them, to keep users safe, and to uphold users’ rights to free speech.

The duties set out in new clauses 3 and 4 will not apply to illegal content, content that is harmful to children or consumer content. That is because illegal content and content that is harmful to children are covered by existing duties in the Bill, and consumer content is already regulated under consumer protection legislation. Companies will also be able to remove any content where they have a legal obligation to do so, or where the user is committing a criminal offence, even if that is not covered in their terms of service.

New clause 5 will require Ofcom to publish guidance to help providers of category 1 services to understand what they need to do to comply with their new duties. That could include guidance on how to make their terms of service clear and easy for users to understand, and how to operate an effective reporting and redress mechanism. The guidance will not prescribe what types of content companies should include in their terms of service, or how they should treat such content. That will be for companies to decide, based on their knowledge of their users, and their brand and commercial incentives, and subject to their other legal obligations.

New clause 6 clarifies terms used in new clauses 3 and 4. It also includes a definition of “Consumer content”, which is excluded from the main duties in new clauses 3 and 4. This covers content that is already regulated by the Competition and Markets Authority and other consumer protection bodies, such as content that breaches the Consumer Protection from Unfair Trading Regulations 2008. These definitions are needed to provide clarity to companies seeking to comply with the duties set out in new clauses 3 and 4.

The remaining amendments to other provisions in the Bill are consequential on the insertion of these new transparency, accountability and free speech duties. They insert references to the new duties in, for example, the provisions about content reporting, enforcement, transparency and reviewing compliance. That will ensure that the duties apply properly to the new measure.

Amendment 30 removes the duty on platforms to include clear and accessible provisions in their terms of service informing users that they have a right of action in court for breach of contract if a platform removes or restricts access to their content in violation of its terms of service. This is so that the duty can be moved to new clause 4, which focuses on ensuring that platforms comply with their terms of service. The replacement duty in new clause 4 will go further than the original duty, in that it will cover suspensions and bans of users as well as restrictions on content.

Amendments 46 and 47 impose a new duty on Ofcom to have regard to the need for it to be clear to providers of category 1 services what they must do to comply with their new duties. These amendments will also require Ofcom to have regard to the extent to which providers of category 1 services are demonstrating, in a transparent and accountable way, how they are complying with their new duties.

Lastly, amendment 95 temporarily exempts video-sharing platforms that are category 1 services from the new terms of service duties, as set out in new clauses 3 and 4, until the Secretary of State agrees that the Online Safety Bill is sufficiently implemented. This approach simultaneously maximises user protections by the temporary continuation of the VSP regime and minimises burdens for services and Ofcom. The changes are central to the Government’s intention to hold companies accountable for their promises. They will protect users in a way that is in line with companies’ terms of service. They are a critical part of the triple shield, which aims to protect adults online. It ensures that users are safe by requiring companies to remove illegal content, enforce their terms of service and provide users with tools to control their online experiences. Equally, these changes prevent arbitrary or random content removal, which helps to protect pluralistic and robust debate online. For those reasons, I hope that Members can support the amendments.

--- Later in debate ---

[Paul Scully]

I will have a go at that, but I am happy to write to the hon. Lady if I do not respond as fully as she wants. Down-ranking content is a moderation action, as she says, but it is not always done just to restrict access to content; there are many reasons why people might want to do it. Through these changes, we are saying that the content is not actually being restricted; it can still be seen if it is searched for or otherwise encountered. That is consistent with the clarification.

[Damian Collins]

This is quite an important point. The hon. Member for Aberdeen North was talking about recommendation systems. If a platform chooses not to amplify content, that is presumably not covered. As long as the content is accessible, someone could search and find it. That does not inhibit a platform’s decision, for policy reasons or whatever, not to actively promote it.

--- Later in debate ---

[Paul Scully]

I will come back to some of the earlier points. At the end of the day, when platforms change their terms and conditions, which they are free to do, they will be judged by their users and indeed the advertisers from whom they make their money. There are market forces—I will use that phrase as well as “commercial imperative”, to get that one in there—that will drive behaviour. It may be the usability of Facebook, or Twitter’s terms and conditions and the approach of its new owner, that will drive those platforms to alternative users. I am old enough to remember Myspace, CompuServe and AOL, which tried to box people into their walled gardens. What happened to them? Only yesterday, someone from Google was saying that the new artificial intelligence chatbot—ChatGPT—may well disrupt Google. These companies, as big as they are, do not have a right to exist. They have to keep innovating. If they get it wrong, then they get it wrong.

[Damian Collins]

Does my hon. Friend agree that this is why the Bill is structured in the way it is? We have a wide range of priority illegal offences that companies have to meet, so it is not down to Elon Musk to determine whether he has a policy on race hate. They have to meet the legal standards set, and that is why it is so important to have that wide range of priority illegal offences. If companies go beyond that and have higher safety standards in their terms of service, that is checked as well. However, a company cannot avoid its obligations simply by changing its terms of service.

[Paul Scully]

My hon. Friend is absolutely right. We are putting in those protections, but we want companies to have due regard to freedom of speech.

I want to clarify a point that my hon. Friend made earlier about guidance to the new accountability, transparency and free speech duties. Companies will be free to set any terms of service that they want to, subject to their other legal obligations. That is related to the conversations that we have just been having. Those duties are there to properly enforce the terms of service, and not to remove content or ban users except in accordance with those terms. There will no platform risk assessments or codes of practices associated with those new duties. Instead, Ofcom will issue guidance on how companies can comply with their duties rather than codes of practice. That will focus on how companies set their terms of service, but companies will not be required to set terms directly for specific types of content or cover risks. I hope that is clear.

To answer the point made by the hon. Member for Pontypridd, I agree with the overall sentiment about how we need to protect freedom of expression.

[Damian Collins]

I want to be clear on my point. My question was not related to how platforms set their terms of service, which is a matter for them and they are held to account for that. If we are now bringing in requirements to say that companies cannot go beyond terms of service or their duties in the Bill if they are going to moderate content, who will oversee that? Will Ofcom have a role in checking whether platforms are over-moderating, as the Minister referred to earlier? In that case, where those duties exist elsewhere in the Bill, we have codes of practice in place to make sure it is clear what companies should and should not do. We do not seem to be doing that with this issue.

[Paul Scully]

Absolutely. We have captured that in other parts of the Bill, but I wanted to make that specific bit clear because I am not sure whether I understood or answered my hon. Friend’s question correctly at the time.

Question put and agreed to.

Clause 20, as amended, accordingly ordered to stand part of the Bill.

Clause 21

Record-keeping and review duties

Amendments made: 32, in clause 21, page 23, line 5, leave out “, 10 or 12” and insert “or 10”.

This amendment is consequential on Amendment 6 (removal of clause 12).

Amendment 33, in clause 21, page 23, line 45, leave out paragraph (c).

This amendment is consequential on Amendment 7 (removal of clause 13).

Amendment 34, in clause 21, page 24, line 6, leave out “section” and insert “sections”.

This amendment is consequential on Amendment 35.

Amendment 35, in clause 21, page 24, line 6, at end insert—

“, (Duty not to act against users except in accordance with terms of service) and (Further duties about terms of service) (duties about terms of service).”—(Paul Scully.)

This amendment ensures that providers have a duty to review compliance with the duties set out in NC3 and NC4 regularly, and after making any significant change to the design or operation of the service.

Question proposed, That the clause, as amended, stand part of the Bill.

--- Later in debate ---

[Kirsty Blackman]

We have seen that just from the people from external organisations who have contacted us about the Bill. The amount of expertise that we do not have that they have brought to the table has significantly improved the debate and hopefully the Bill. Even prior to the consultations that have happened, that encouraged the Minister to make the Bill better. Surely that is why the pre-legislative scrutiny Committee looked at the Bill—in order to improve it and to get expert advice. I still think that having specific access to expertise in order to analyse the transparency report has not been covered adequately.

[Damian Collins]

Annual transparency reporting is an important part of how the system will work. Transparency is one of the most important aspects of how the Online Safety Bill works, because without it companies can hide behind the transparency reports they produce at the moment, which give no transparency at all. For example, Facebook and YouTube report annually that their AI finds 95% of the hate speech they remove, but Frances Haugen said that they removed only 5% of the hate speech. So the transparency report means that they remove 95% of 5%, and that is one of the fundamental problems. The Bill gives the regulator the power to know, and the regulator then has to make informed decisions based on the information it has access to.

--- Later in debate ---

[Charlotte Nichols]

As much as I am keen on the idea of Ofcom special agents conceptually, my concern on the transparency front is that, to appoint a special agent and send them in to look at the data, Ofcom would have to have cause to believe that there was an issue of concern with the data, whereas if that data is more transparently available to the research community, they can then proactively identify things that they can flag to Ofcom as a concern. Without that, we are relying on an annual cycle of Ofcom being able to intervene only when they have a concern, rather than the research community, which is much better placed to make that determination, being able to keep a watching brief on the company.

[Damian Collins]

That concern would be triggered by Ofcom discovering things as a consequence of user complaint. Although Ofcom is not a complaint resolution company, users can complain to it. Independent academics and researchers may produce studies and reports highlighting problems at any time, so Ofcom does not have to wait through an annual cycle of transparency reporting. At any time, Ofcom can say, “We want to have a deeper look at this problem.” It could be something Ofcom or someone else has discovered, and Ofcom can either research that itself or appoint an outside expert.

As the hon. Member for Warrington North mentioned, very sensitive information might become apparent through the transparency reporting that one might not necessarily wish to make public because it requires further investigation and could highlight a particular flaw that could be exploited by bad actors. I would hope and expect, as I think we all would, that we would have the routine publication of transparency reporting to give people assurance that the platforms are meeting their obligations. Indeed, if Ofcom were to intervene against a platform, it would probably use information gathered and received to provide the rationale for why a fine has been issued or another intervention has been made. I am sure that Ofcom will draw all the time on information gathered through transparency reporting and, where relevant, share it.

[Paul Scully]

This has been a helpful debate. Everyone was right that transparency must be and is at the heart of the Bill. From when we were talking earlier today about how risk assessments and terms of service must be accessible to all, through to this transparency reporting section, it is important that we hold companies to account and that the reports play a key role in allowing users, Ofcom and civil society, including those in academia, to understand the steps that companies are taking to protect users.

Under clause 65, category 1 services, category 2A search services and category 2B user-to-user services need to publish transparency reports annually in accordance with the transparency report notice from Ofcom. That relates to the points about commerciality that my hon. Friend the Member for Folkestone and Hythe talked about. Ofcom will set out what information is required from companies in their notice, which will also specify the format, manner and deadline for the information to be provided to Ofcom. Clearly, it would not be proportionate to require every service provider within the scope of the overall regulatory framework to produce a transparency report—it is also important that we deal with capacity and proportionality—but those category threshold conditions will ensure that the framework is flexible and future-proofed.