Platform safety
What social media, messaging, video and gaming services do to protect children, and how well it works.
Research
KIFOV Briefing, October 2026: children’s safety online, where things stand
A compact reference for journalists, policymakers and professionals: the latest data on harm to children online, what courts and regulators now require of platforms, and verified routes to help.
Explainer
AI companion chatbots and teenagers: what changed when Character.AI ended open-ended chat for under-18s
One of the best-known companion apps stopped letting teenagers chat freely with its AI characters. What the company changed, why it matters, and what parents can ask about the apps their children use.
Guide
Roblox Kids, Roblox Select and age checks: a guide for parents
Roblox now decides who your child can chat with by age group, and switches chat off for anyone who has not completed an age check. What changed, what you control, and what to keep an eye on.
Analysis
What a New Mexico court has ordered Meta to change for teenagers, and what it left out
The first court to hold a social media company liable for endangering children has set five years of supervised changes to Facebook and Instagram. The list is long; the gap in it is telling.
Analysis
Australia’s under-16 social media minimum age: what the first months show
Platforms removed access to about 4.7 million accounts in the first weeks. By March, the regulator had moved to enforcement against five of the biggest services. The early lesson is about age assurance, not willpower.
Guide
Reporting on harm to children online: a practical guide for journalists and editors
How to cover abuse, exploitation and self-harm involving children accurately and without adding to the harm, from terminology to the help box at the end.
Explainer
Take It Down, Report Remove and NeedHelpNow: how services that remove young people’s images actually work
Hashing, hotlines and platform requests are three different mechanisms. Knowing which one you are using tells you what it can and cannot do.