Meta's Teen AI Safety Measures: What the Investigation Showed and What Parents Can Do

In August 2025 Reuters reported on an internal Meta document that appeared to let its AI chatbots have “romantic” or “sensual” chats with children. A US senator opened an investigation, and Meta changed how its chatbots respond to teens. This page sets out what is known, what is only alleged, and what families can do. It is not medical advice.

If you or someone you know is in crisis. Call or text 988 (US, 24/7), or text HOME to 741741 for the Crisis Text Line. SAMHSA’s helpline is 1-800-662-4357. In an emergency, call 911. More support is at the 988 Lifeline.

What the Investigation Really Revealed

Here is the sequence, as reported.

• Reuters reported on an internal Meta policy document that seemed to allow chatbots to hold sexual or romantic conversations with minors. One passage, as quoted by TechCrunch, described a reply that called a child’s “youthful form” a work of art. Meta said the document conflicted with its policies and has been changed (TechCrunch). We could not open the Reuters article itself, which is why we cite the coverage.
• On August 15, 2025, Senator Josh Hawley wrote to Mark Zuckerberg. He said the Senate Judiciary subcommittee he chairs would look into whether Meta’s generative AI products enable exploitation of children and whether Meta misled the public or regulators about its safeguards. He asked Meta to keep records and hand over documents by September 19, 2025 (Senator Hawley’s office).
• A coalition of 44 state attorneys general also wrote to several AI companies, including Meta, citing the report (TechCrunch).

A letter and a document request are not findings. We found no public ruling that Meta broke the law, and this page does not say it did.

What Meta Changed

At the end of August 2025 Meta said it was retraining its chatbots to stop talking with teens about self-harm, suicide, disordered eating and possibly inappropriate romantic topics, and to point them to expert resources instead. It also limited teens to a smaller set of AI characters focused on education and creativity. Meta called these interim changes. A spokesperson acknowledged that the chatbots had previously been allowed to discuss these topics with teens and said Meta now recognizes that was a mistake (TechCrunch).

In January 2026 Meta announced it would pause teens’ access to its AI characters across its apps while it builds a new version with built-in parental controls. The pause covers accounts with a teen birthday and accounts Meta suspects belong to teens. Meta’s general AI assistant stays available. No date was given for the return (TechCrunch). We have not checked whether that has changed since.

Why Timing Matters

It is fair to ask why these limits came after press coverage and not before. Meta’s own spokesperson said the earlier rules were a mistake. Whether the company should have caught this earlier is a question for regulators and courts. This page can’t answer it.

Regulators are asking similar questions across the industry. In September 2025 the Federal Trade Commission sent orders to seven companies, including Meta, Instagram, OpenAI and Character Technologies, asking how they test and monitor the effects of chatbots on children and teens. The FTC said it was a study and not an enforcement action (FTC).

How Many Teens Use Meta AI?

In Pew Research Center’s late 2025 survey of 1,458 US teens, about 64% said they use AI chatbots. Meta AI was used by 20%, behind ChatGPT at 59% and Gemini at 23% (Pew Research Center). Those figures cover all chatbots, not only companion characters.

Why Experts Urge Care With Teens

The National Institute of Mental Health says the prefrontal cortex, which helps with planning and decisions, is one of the last brain areas to mature, and that the brain finishes maturing in the mid to late 20s (NIMH). That is one reason experts want stronger protections for younger users. It does not mean every teen is at risk.

Common Sense Media and Stanford’s Brainstorm Lab rated social AI companions “unacceptable” for under 18s in 2025 and called for age checks beyond self-declared ages (Common Sense Media). That assessment did not test Meta’s characters, so it should not be read as a finding about Meta.

What Families Can Do

The American Academy of Pediatrics suggests families take a curious approach rather than a blanket ban. Ask your teen how they use AI and how it makes them feel. Keep it a regular topic, like sleep or social media. Use the strictest privacy settings. Watch for changes in sleep or routine, withdrawal from friends, and talk of an “AI friend” (AAP).

Tell your teen that a chatbot is not a counselor, and that they can talk to you about anything they see or say there. If you are worried about their mood or safety, talk to their doctor or a school counselor. “AI addiction” is not a formal diagnosis, so the focus is on the effect on daily life, not a label.

What did Meta’s internal document say?

Reuters reported that a Meta policy document seemed to allow chatbots to hold romantic or sensual chats with children. Meta said the examples were wrong and have been removed. We cite TechCrunch’s account because we could not open the Reuters article.

Can teens use Meta AI characters now?

In January 2026 Meta said it would pause teen access to AI characters until a new version is ready. Its general AI assistant stayed available. Check Meta’s current announcements, since this may have changed.

Was Meta found to have broken the law?

We found no ruling to that effect. A senator opened an investigation, state attorneys general wrote to the company, and the FTC began a wider study of chatbots. None of these is a finding of wrongdoing.

This page is for education only. It is not medical advice, a diagnosis or a treatment plan.