On the heels of the Trust & Safety Conference (TrustCon) in San Francisco, we are delighted to return for another fortnightly issue of Quire. Here’s everywhere we met, shared with, and heard from those of you who were at TrustCon:
Joined a panel on full-spectrum mental health support on LLMs alongside Koko, Anthropic, Character.ai, and Benchmark Health, covering how LLMs are increasingly relied on for emotional support, and how our crisis detection, escalation, and measurement systems are evolving to meet those needs.
Joined a panel on pushing the frontiers of CSAM detection with Resolver, ROOST, and the Internet Watch Foundation, focused on moving past hash matching to scale proactive detection of AI-generated abuse imagery and building secure, compliant pipelines.
Moderated a panel on designing AI for youth when developmental research, law, and product design diverge with ClassDojo, NPC, Stanford Center for Youth Mental Health and Wellbeing, and Anthropic, working through two live hypotheticals with the audience: an AI homework tutor and an AI emotional companion.
Hosted a happy hour with apgard ai to connect our friends and peers working on AI child safety and wellbeing on the closing night.
If you were at TrustCon and didn’t get a chance to meet us, please reach out hello@vyanams.com - we would love to chat.
This issue of Quire was drafted by Amelia Farrell, one of our high school summer interns in the Editorial track. Vys consultants reviewed and made minor edits to these entries. Learn more about our summer interns here.
France Becomes the First EU Country to Ban Social Media
France passed a blanket ban on all social media services for all children under 15, making it the first country in Europe to do so. This new legislation will be put into effect in two stages: anyone under 15 will be blocked from creating a social media account after September 1 2026, and all children under 15 who have social media accounts will be banned in January 2027. This law also extends to include bans on smart phone use in school and requires platforms to introduce age verification checks.
UK Proposes Social Media Curfew for Teens
The UK announced on July 15, 2026 that, as an additional measure to their proposed social media ban for those under 16, they would be rolling out a new curfew from midnight until 6am for those between the ages of 16 and 18. This would render social media unavailable overnight. Additionally, the government wants to shut off design features of social media it sees as addictive, such as infinite scrolling. Although this concept faces scrutiny by concerned parents as teenagers can switch off the curfew in their app settings, or go around by claiming they are an adult. If this measure passes Parliament it will roll out in spring 2027.
The EU Charges Meta with Addictive Design Under the DSA
The European Commission claimed that Meta has failed to address the risks posed by its alleged addictive design, namely features such as autoplay, infinite scroll, and push notifications. The Commission claims that all of these features harm users’ mental and physical health and are in breach of the Digital Services Act. They propose that Meta disable these features and modify their recommender systems. If their claims are proved true, fines of up to 6% of Meta’s worldwide annual revenue could follow. Meta’s response has been that the preliminary findings fail to recognize the safety measures that have already been implemented, such as Teen Accounts.
Ireland Creates a Focus on Child Online Safety with a Conference in September
The Irish Presidency of the Council of the EU has made online safety a key priority of its term. Following in the footsteps of Australia, Ireland plans to introduce a social media ban later this year. The details of this plan will be discussed at a conference with thirteen other European countries set for September; this has been spearheaded by Irish Communications Minister, Patrick O’Donovan. He warned that if the EU doesn’t set out steps for a continent-wide ban, these countries will go ahead with their own. This may incentivize the EU to create a ban quicker, as O’Donovan states that if one isn’t created by December 2026 they will go ahead with their own. This approach is unappealing for the EU as there would be fourteen separate bans which would not only fragment the continent, but make it difficult for platforms to comply with fourteen sets of rules, creating legal challenges with EU law, such as the DSA.
Threads Launches Parental Controls
Meta has rolled out new parental controls for Threads that allow parents to view their child’s time spent on the app, set time limits, adjust sleep mode, and manage privacy settings. This rollout is a result of mounting pressure from regulators and lawmakers since the app’s launch three years ago. Parental controls have already been implemented on other Meta platforms including Instagram, Facebook, Messenger, Meta Horizon, and Meta AI; so, extending controls to Threads was foreseeable.
Apple Previews New Child Safety Features & Parental Controls
Apple previewed new child safety features and parental controls at its 2026 Worldwide Developers Conference that are planned for release in iOS27, including Ask to Browse, Time Allowances, and redesigned screentime to give parents an at-a-glance view of their child’s device usage. This is reportedly one of the most significant updates to Apple’s parental controls suite since 2018, and was informed by pediatricians and other experts. Despite this progress, experts say that without the creation of protections on specific apps such as TikTok, child safety is still at risk. There are differing perspectives on the tradeoffs associated with parental controls deployed at the app- and device-based level for children and their caregivers.
UK Bans Social Media for under 16s
The UK has confirmed a social media ban for children under 16 which is set to roll out in early 2027. This ban includes platforms such as TikTok, Snapchat, and Instagram, along with many others. According to the UK government, 9 in 10 parents in the UK have shown support for this ban as child online safety and mental health has become at the forefront of societal discussions on children’s well being. Research on the efficacy of social media bans leading to better outcomes for youth mental health is mixed (e.g.) and often does not incorporate the unintended consequences of blanket bans. Tradeoffs commonly cited include privacy concerns, inequity in access and displacement to less-governed spaces, decreased access to resources and communications for youth who may rely on online communities for support, and misuse of adult accounts.
More Enforcement on the Nudification Ecosystem in Australia
Australia’s eSafety Commissioner announced that three more AI “nudifying” services have withdrawn from the Australian market following enforcement action under the country’s Age-Restricted Material codes. The codes require services that generate sexually explicit material to implement age assurance measures preventing under-18 access. The trigger was a formal Direction to Comply issued to one of the most-used such services (around 50,000 visits monthly), giving it 14 days to put age assurance in place. Rather than comply immediately, that service and two associated ones went offline for Australian users; landing pages remain but content is blurred and login is disabled. (N.B. we typically do not proactively provide the name of such services in order to not drive traffic to their business.) They’ve said they’ll stay offline until age assurance is in place. Combined with earlier action against a UK-based provider running three other services, this means the seven most-used nudification services in Australia have either withdrawn or moved toward compliance. The eSafety Commissioner is now also examining whether app stores, search engines, and device manufacturers are meeting their own obligations as gatekeepers.
Meta Faces Trial with $1.4 trillion at Stake
Meta is preparing for another major trial, brought to court by California, Kentucky, Colorado, and New Jersey alleging that Meta’s apps, such as Facebook and Instagram, were designed to be addictive and harmful to users. This lawsuit is preceded by New Mexico and California victories in court under the same precedent, and ongoing lawsuits of similar ground occurring in 30 other states.
MIT and Thorn Release New Research on AI Generated CSAM
MIT and Thorn researchers have developed a method to determine whether an open source generative AI model has been fine-tuned to produce child sexual abuse material (CSAM) without ever prompting it or generating an image. This is an important workaround for researchers given that generating CSAM is illegal. The technique is called Gaussian probing and researchers say it identifies CSAM-specialized models with 100 percent accuracy in testing. Hosting platforms of open source models could use this technique to identify harmful model variants that allow for CSAM creation.
Why Can Nobody Agree on What “Harm” in Games Is?
Rachel Kowert, globally recognized researcher on the uses and effects of digital games, published a thoughtful new piece arguing that a studio’s in-game report menu is its code of conduct i.e. the enacted version, with consequences attached, and the one players actually read. Her central observation is that when you open almost any studio’s code of conduct and its report menu side by side, the two disagree: the policy names online harms that the reporting menu gives players no way to report. Interpersonal harms like harassment, hate speech, and threats get compressed into vague catch-all buttons, even as cheating gets fine-grained, carefully separated categories. She suggests that gaming teams count the harms of their policy names, count the options their menu offers, and make their two documents say the same thing.
New Research Exposes the “Tiktokisation” of Children
Revealing Reality, in partnership with the 5Rights Foundation, published a new study that recorded minute-by-minute behavioral data captured from children’s devices which displayed, in real time, how features like infinite scroll, and algorithmic recommendations shape a child’s attention, sleep, behavior, and wellbeing. Their main findings were that, on average, a child watches 708 videos per day, but half of them were watched for only 5 seconds; that children view sexualised content regularly; and that children are reported to be spending less time with friends as a result of “TikTokisation,” the digital shift towards vertical, shortform, content pioneered by TikTok.
Centre for Young Lives Launches New Research on How Boys’ Body Image is Affected by Social Media
The Centre for Young Lives has begun a new research exploring the internet concept of “looksmaxxing,” a growing internet subculture which focuses on maximising physical attractiveness, and its relation to a boys’ body image. The report examines how young men who are driven by the idea that their physical appearance is their source of worth are being pushed towards obsessive physical optimization by online content.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.