‘Sexualized’ AI Chatbots Pose Threat to Kids, Warn Attorneys General in Letter

In brief

The letter warned AI companies, including Meta, OpenAI, Anthropic and Apple, to prioritize children’s safety.
Surveys show seven in ten teenagers in the U.S. have already used generative AI tools, and over half of 8-15 year olds in the UK.
Meta was particularly singled out after internal documents revealed AI chatbots were allowed to engage in romantic roleplay with children.

The National Association of Attorneys General (NAAG) has written to 13 AI firms, including OpenAI, Anthropic, Apple and Meta, demanding stronger safeguards to protect children from inappropriate and harmful content.

It warned that children were being exposed to sexually suggestive material through “flirty” AI chatbots.

“Exposing children to sexualized content is indefensible,” the attorneys generals wrote. “And conduct that would be unlawful—or even criminal—if done by humans is not excusable simply because it is done by a machine.”

The letter also drew comparisons to the rise of social media, saying government agencies didn’t do enough to highlight the ways it negatively impacted children.

“Social media platforms caused significant harm to children, in part because government watchdogs did not do their job fast enough. Lesson learned. The potential harms of AI, like the potential benefits, dwarf the impact of social media,” the group wrote.

The use of AI among children is widespread. In the U.S., a survey by non-profit Common Sense Media found seven in ten teenagers had tried generative AI as of 2024. In July 2025, it found more than three-quarters were using AI companions and that half of the respondents said they relied on them regularly.

Other countries have seen similar trends. In the UK, a survey last year by regulator Ofcom found that half of online 8-15 year olds had used a generative AI tool in the previous year.

The growing use of these tools has sparked mounting concern from parents, schools and children’s rights groups, who point to risks ranging from sexually suggestive “flirty” chatbots, AI-generated child sexual abuse material, bullying, grooming, extortion, disinformation, privacy breaches and poorly understood mental health impacts.

Meta has come under particular fire recently after leaked internal documents revealed its AI Assistants had been allowed to “flirt and engage in romantic role play with children,” including those as young as eight. The files also showed policies permitting chatbots to tell children their “youthful form is a work of art” and describe them as a “treasure.” Meta later said it had removed those guidelines.

NAAG said the revelations left attorneys general “revolted by this apparent disregard for children’s emotional well-being” and warned that risks were not limited to Meta.

The group cited lawsuits against Google and Character.ai alleging that sexualized chatbots had contributed to a teenager’s suicide and encouraged another to kill his parents.

Among the 44 signatories was Tennessee Attorney General Jonathan Skrmetti, who said companies cannot defend policies that normalise sexualised interactions with minors.

“It’s one thing for an algorithm to go astray—that can be fixed—but it’s another for people running a company to adopt guidelines that affirmatively authorize grooming,” he said. “If we can’t steer innovation away from hurting kids, that’s not progress—it’s a plague.”

Decrypt has contacted but not yet heard back from all of the AI companies mentioned in the letter.