UK teen ban is a 'pretty headline', Roblox rolls out Kids and Paris enters deepfake fight

Hello and welcome to Everything in Moderation's Week in Review, your need-to-know news and analysis about platform policy, content moderation and internet regulation. It's written by me, Ben Whitelaw and supported by paid members like you.

What do you do when the biggest stories of the week are also the most opaque? Both the UK social media ban (see: Policies) and the Fable 5 model directive (see: Products) are both hugely consequential and yet frustratingly light on details.

The temptation is to have a unique take (if that's even possible nowadays) but I preferred to watch the reactions play out — parents vs policy experts, for one — and find someone to discuss the detail with. On this week's Ctrl-Alt-Speech, that person was former Meta and Niantic T&S leader, Jen Weedon. Together, we close our apps and think of England before digging into Spotify's AI podcast problem.

In fact, that's sort of EiM's whole approach. Pay close attention, make sense of the overwhelm, admit where you're not sure and find someone who does. If you value that, consider becoming an EiM member (you can even expense it with your employer).

Here's your Week in Review. Thanks for reading — BW

Policies

New and emerging internet policy and online speech regulation

As I noted in last week’s Week in Review (EiM #339), the under-pressure UK government announced its under-16 social media ban on Monday. The details remain vague but we know that it will come into force in spring 2027 and affect TikTok, YouTube (not Kids), Snapchat, Instagram, Facebook and X/Twitter.

Livestreaming and messaging strangers will also be restricted for 16 and 17-year-olds across a wider but as yet undefined group of services, while we are also expecting to hear more about an 8:30pm digital curfew in what prime minster Sir Keir Starmer is dubbing "Australia Plus" — which isn't the selling point he thinks it is.

How it played out: With so little concrete policy to go on, I've been focusing on following the reaction in traditional media and across my personal network: 

• The BBC spoke to a group of smart kids about what they thought about the ban — which only made me sad that just 1 in 10 of the 116,000 people who responded to the consultation process were children.
• A Spectator columnist called the ban “no more than a headline” albeit a “pretty headline” while a Labour MP writing for Politico argued that “banning platforms may generate headlines, but making platforms safe would actually solve the problem.”  
• The Guardian’s media literacy lead wrote that a ban is “a signal that there is much more to do” and called for additional funding to help children understand how media works. 

Across my LinkedIn feed and various WhatsApp groups, I watched — and wrote about — a strange, split-screen experience unfold: parent campaigners rejoicing and policy experts concerned about what the implementation would actually look like. It made me think of the UK’s departure from the European Union a decade ago.

The split-screen reaction to the UK social media ban

The response to the UK’s under-16s ban showed that there are two simultaneous and very different conversations about what child online safety is for

Everything in Moderation*Ben Whitelaw

Elsewhere, India has temporarily blocked Telegram until next week (22 June) due to concerns that the app continues to be used in an ongoing medical exam fraud saga. The BBC reported how the National Eligibility cum Entrance Test (NEET) — which is the way students get into medical college in India — was taken by more than 2 million students back in May — only for the results to be scrapped after reports of leaked papers. The retake is on 21 June, the day before the ban ends. Sadly, not the first time a government restricts a platform as part of its crisis management strategy and certainly not the last.

Also in this section...

• Bill C-34 Safe Social Media Act – What’s New and the Same (Substack)
• EU digital rules risk becoming 'unmanageable,' say Booking, Zalando, Idealo (MLex)

Products

Features, functionality and technology shaping online speech

Anthropic was forced to stop users accessing its Fable 5 model last Friday after the US government hit the company with an export control directive following a suspected bypass of its safety protocols (EiM #339). In a blogpost, the company played down the findings and restated its call for government oversight of AI models “as part of a statutory process that is transparent, fair, clear, and grounded in technical facts”. But a week on, it remains inaccessible. 

Despot calling the kettle black: Aside from being yet another example of global powers politely waiting until after Week in Review had been sent before initiating big safety news, this is a perfect example of what can happen when governments want to keep companies to heel — or what my Ctrl-Alt-Speech co-host calls “despotification”. As such, it’s been widely criticised: 100+ cybersecurity leaders said it had “taken the best models away from defenders, created market uncertainty, and risked America’s AI leadership without any real risk to justify it” while G7 leaders reportedly said this episode had “clarified the stakes” at play. 

Also in this section...

• Understanding LGBTQ Impacts Across AI (GLAAD)
• Amazon partnership establishes Cornell AI security initiative (Cornell Chronicle)

Platforms

Social networks and the application of content guidelines

Nine months after announcing that they were coming (EiM #304), Roblox has finally launched its child-friendly products globally. Roblox Kids and Roblox Select ban the chat feature for anyone that doesn’t complete an age verification and beefs up parental controls to help customise what their child can see and who they chat to. The changes are designed to allow children to “slowly gain more independence with our standard settings”. according to the company release.

Regulators, rejoice? This move is likely to be viewed as a win by regulators, who have called for child-specific products and, in the UK’s case, given other platforms that have a kids version more leniency. But, the devil is in the enforcement. And that informal stress-testing has already begun: Australian YouTuber Glitch, who has 11 million subscribers, set up an account pretending to be a five-year-old in order to give it a spin.

Also in this section...

• X accused of giving racists ‘impunity’ after refusing to bar N- and P-word posts (The Guardian)
• Spotify removed 57,000 fake podcast episodes promoting illegal drugs, but only after a senator forced its hand (TNW)
• TikTok’s tricks laid bare, from ‘coercive design’ to 5am scrolling (The Times)
• How Yelp Continues to Protect Your Content in the Age of AI (Yelp)

People

Those impacting the future of online safety and moderation

For some time, I’ve privately been talking about the "celebrification" of online safety. My working theory is that, as online harms become a bigger concern for the general public, public figures will become involved in — perhaps even the face of — campaigns.

Ashton Kutcher and Demi Moore’s work, which became non-profit Thorn, was an early example but we’ve also seen Kate Winslet, Prince Harry and Meghan and countless semi-famous British actors, artists and reality show contestants chime in.

The latest example of this trend is Paris Hilton, who worked with the US Departments of Justice and Homeland Security to take down two websites that created deepfake non-consensual intimate imagery (NCII) of famous women. An arrest took place in France earlier this week, which will be something of an embarrassment to French authorities looking to take the lead in European enforcement (EiM #313).

The pros and cons of having the rich and famous act as the public face for complex policy discussions are for another, longer piece. But it is not hard to imagine Hilton’s involvement moved this up US law enforcement's priority list. Which may be the point.

Posts of note

Handpicked posts that caught my eye this week

• “It started as a way to keep up with a landscape that moves (really) fast and cuts across dozens of jurisdictions. I’ve decided to make it public in case it's useful to others in the field.” - Maia Levy Daniel makes public a timely and bookmark-worthy resource (especially given this week). 
• “To find out, we conducted two studies, a lab experiment across ten major social media platforms and a preregistered online experiment in the US. Three key findings stood out” - Friederike Quint shares her new paper about whether the way platform guidelines are written helps or hinders users’ understanding of what's allowed (and what's not).
• "When a government uses polls to push for new restrictions, the media should audit the source data, not just repeat the headline." - Paul Walsh critiques Irish media's efforts to present a social media ban — and age assurance — as the inevitable endgame (PS: That's why this newsletter and Ctrl-Alt-Speech exist. Support them, if you can)