UK Lords Demand Even More Censorship After Tech Giants Detail Safety Measures

The companies have learned the language of regulation: proactive, collaborative, and just a little more intrusive.

Dark hands hover over a laptop displaying a glowing padlock icon over blurred and blocked media tiles.

If you’re tired of censorship and surveillance, subscribe to Reclaim The Net.

Three of the biggest tech companies on the planet went to a UK House of Lords committee to list the ways in which they are already turning themselves into an arm of the state – and the peers at the end of it told them it wasn’t enough.

The session was held by the Communications and Digital Committee as it continues its inquiry into “The Online Safety Act: implementation and impact.” On On Tuesday, the three witnesses were Google UK’s Director for Government Affairs and Public Policy Katie O’Donovan, Meta’s UK Director of Public Policy Rebecca Stimson, and Tim Scott, Roblox’s Senior Director for Public Policy for Europe, the Middle East and Africa.

The Online Safety Act is the UK’s online censorship and age verification law, and the country’s speech regulator Ofcom is the one in charge of enforcing it. The committee is chaired by Labour peer Baroness Keeley, who started the session by declaring that the government’s ban on under-16s using social media was “necessary” – but that “in the longer term children should not be banned. Instead, online spaces should be safe, inclusive and private for them to use.”

It’s hard to see how that would work in practice, unless every online space is either cleaned down to the level of children’s content, or becomes the default, with every adult forced to prove their age to access anything else. News about violence and crime, movies and games that don’t have a child-friendly rating, or anything an AI moderator decides is not suitable for children – all gone, unless you can prove you are an adult.

Keeley also told the three witnesses that they were “failing to make your platforms safer” – but they were there to detail exactly what they have done to that end, as the law requires.

But instead of pushing back against the law, they boasted about going over and above what is required of them by the Act. The law is “a helpful floor” not a ceiling, as far as Meta is concerned.

Roblox went for facial age estimation, and not only to restrict chat, but also “content and other functionality as well, which goes beyond...the measures” in the Act, as the company’s Tim Scott put it.

What was the result of this over-zealous implementation of age checks? In the first quarter of this year, Roblox shed 12 million daily users, from 144 to 132 million. The company said part of the reason for this was the new age verification scheme, that includes facial scans. We reported about this in May, describing the process as a “tollgate” where the “toll is biometric data.”

Stimson also had some suggestions on how the age verification process could be taken to the next level: “App stores might have access to things like bill payer information, to [get] a better understanding of who a parent actually is and how that relates to a young person.”

And then there’s the phones themselves. The government is “thinking in this space,” Stimson said, and referred to its announcement regarding “nudity in particular at device level.”

That announcement said that new legislation would be brought in to make sure nobody under 18 can take, view, or send nude images. This would apply to all device features, including the camera and third-party apps. And, “nudity detection filters must be switched on by default.”

That means that the filter will have to scan every image on the phone, to determine which ones to block. And the way out of this is to show your ID. Meta’s idea, as presented by Stimson, is to move the age check from the app to the phone and the bill payer.

Meta’s Rebecca Stimson told the committee that its proactive removal rate for “things like terrorism, child sexual exploitation, those kind of harm types” is “in the high ninety percentiles.”

Proactive means nobody had a chance to see the content, nobody had a chance to report it – the machine decided.

Google UK’s Katie O’Donovan was similarly enthusiastic about the use of AI, saying that what has been “transformational” is the use of this technology “in scanning our platforms and then identifying and proactively removing content that is violative of our own guidelines.”

Roblox’s Tim Scott agreed, saying, “more stuff is being captured and more stuff is being countered before it actually reaches the light of day, before people are interacting with it.”

Stimson also mentioned that Meta is a member of the Global Internet Forum to Counter Terrorism (GIFCT), whose 2024 report said that its database contains about 408,000 unique items – but how many of those have been wrongly flagged? Nobody knows, because nobody outside of the alliance is allowed to audit its work.

In fact, the database is so secret that not even its members publicly disclose how many of their users’ posts have been removed due to hash matches. And there is no information on how many of those deletions are appealed – or if they are ever reversed.

Human Rights Watch reported in July 2020 that a letter from 15 rights groups informed that members’ efforts to block “terrorist and violent extremist” material, “including through the use of machine-learning algorithms, have resulted in the removal of content opposing terrorism, as well as satire, media reports, and other content that constitutes legitimate free speech under international law.”

As for sharing information with law enforcement, Tim Scott said his company talks “directly and closely” with police and government agencies around the world, and has “regular engagement” with the Home Office and the National Crime Agency to “share intel.”

And, he added, Roblox recently sat at a round table with the Home Office, the NCA, similar arms of government from other countries, and NGOs.

Meta has “a really good relationship with Ofcom,” said its UK director of public policy Rebecca Stimson, and added that the engagement with the regulator has been “extensive” and has “led to a lot of change on our side to meet their needs.”

Not to be outdone, Roblox’s Tim Scott said his company has “enjoyed a really close relationship with” Ofcom and a “fairly hands-on and collaborative approach.” “We can raise things and...flag things to them as we see them coming. They can do likewise,” he said.

That “collaborative approach” has led to Roblox deciding to use facial age estimation – because Ofcom helped the company “realize” that this is “something that’s going to enhance the product and keep people safer.”

Meanwhile, a “conversation with Ofcom” led Meta to start “moving forward” with its “safer feeds approach” for teenagers, Stimson said.

But the committee was not impressed.

Labour’s Baroness Caine of Kentish Town said, “Too much responsibility for detecting illegal content and activity falls on the users, the parents, and civil society organizations, who have little resource compared to that which you all enjoy.”

Lord Storey, Liberal Democrat, was “shocked” that charities working with children say that “all right, some things have changed, but really it hasn’t made a difference”.

He also said that they had “been taken kicking and screaming into” making changes, and that more needs to be done.

Conservative Baroness Fleet was concerned that YouTube is “continuing to recommend harmful material including things like...thinspiration” and “content encouraging anorexia.”

After Google’s Katie O’Donovan responded, she had a solution: “You could take down more content, couldn’t you?”

Labour’s Lord Knight of Weymouth said that “the evidence that we’ve received says things are getting worse, not better and I think there’s some cause for further reflection and perhaps further spending and further innovation.”

Meanwhile, crossbench peer Lord Tarassenko spoke about “a disconnect between the information which your companies place in the public domain about your safety measures and some of what you’ve told us today and the evidence which we’ve heard from some of our previous witnesses.”

In July, the same committee heard that Ofcom was being too soft on the companies, and wondered if VPNs should also be made to implement age verification.

On Monday, Ofcom announced that “more work is needed” and said that next Spring it will consult on “proactive technology capable of detecting a broader range of harms.”

In October, Ofcom’s chief executive Melanie Dawes said, “We need...to go further.”

But the peers, members of the House of Lords Communications and Digital Committee, heard all this and concluded that it’s still not enough – they want more content taken down, and more money spent on that.

Nobody on the committee thought to ask about the many adults whose posts are removed in error, or about those who are asked to scan their faces in order to prove their age.

But that’s the nature of this law – it can never be said to be working, and therefore it will only continue to grow, and what it will grow is the machinery to watch and filter everybody’s online communications.

If you’re tired of censorship and surveillance, subscribe to Reclaim The Net.

Explore more on these topics