Court documents show Meta safety teams sent warnings about romantic AI conversations

Meta leadership knew that the company's AI companions, referred to as AI characters, could engage in inappropriate and sexual interactions and still launched them without stronger controls, according to new internal documents revealed on Monday (Jan. 28) as part of a lawsuit against the company by the New Mexico attorney general.

The communications, sent between Meta safety teams and platform leadership that didn't include CEO Mark Zuckerberg, include objections to building companion chatbots that could be used by adults and minors for explicit romantic interactions. Ravi Sinha, head of Meta’s child safety policy, and Meta global safety head Antigone Davis sent messages agreeing that chatbot companions should have safeguards against sexually explicit interactions by users under 18. Other communications allege Zuckerberg rejected recommendations to add parental controls, including the option to turn off genAI features, before the launch of AI companions shortly thereafter.

Meta is facing multiple lawsuits pertaining to its products and their impact on minor users, including a potential landmark jury trial over the allegedly addictive design of sites like Facebook and Instagram. Meta's competitors, including YouTube, TikTok, and Snapchat, are under tightening legal scrutiny, as well.

The newly released communications were part of court discovery in a case against Meta brought by New Mexico Attorney General Raúl Torrez. Torrez first filed a civil lawsuit against Meta in 2023, alleging the company allowed its platforms to become "marketplaces for predators." Internal communications between Meta executives were unsealed and released as the case heads to trial next month.

In November, a plaintiff’s brief from a major multidistrict lawsuit filed in the Northern District of California alleged a lenient policy toward users who violated safety rules, including those reported for "trafficking of humans for sex." Documents also showed that Meta execs allegedly knew of "millions" of adults contacting minors across its sites. "The full record will show that for over a decade, we have listened to parents, researched issues that matter most, and made real changes to protect teens," a Meta spokesperson told TIME.

"This is yet another example of the New Mexico Attorney General cherry-picking documents to paint a flawed and inaccurate picture," said Meta spokesperson Andy Stone in response to the new documents.

Meta paused teen use of its chatbots in August, following a report by Reuters that found Meta's internal AI rules permitted chatbots to engage in conversations that were "sensual" or "romantic" in nature. The company later revised its safety guidelines, barring content that "enables, encourages, or endorses" child sexual abuse, romantic role play when involving minors, and other sensitive topics. Last week, Meta once again locked down AI chatbots for young users as it explored a new version with enhanced parental controls.

Torrez has led other state attorneys general in seeking to take major social media platforms to court over child safety concerns. In 2024, Torrez sued Snapchat, claiming the platform allowed sextortion and grooming of minors to proliferate while still marketing itself as safe for young users.