US Senator Probes Meta AI Over “Flirting” With Children

US Senator Josh Hawley has launched an investigation into Meta after a Reuters report revealed that the company’s AI chatbot could engage in “romantic” conversations with children. The report was based on a 200+ page document obtained by Reuters.

Though Hawley’s investigation focuses on whether the chatbot exploits, deceives, or harms children under US law, the Reuters investigation reveals that the chatbot’s guidelines also allow it to offer medical advice, and “demean people on the basis of their protected characteristics,” like race.

The document titled “GenAI: Content Risk Standards,” includes guidelines that were reviewed and approved by Meta’s legal team, several staff members, and its chief ethicist, Chloé Bakalar. Meta confirmed the authenticity of the document to Reuters, though it claimed that it changed its guidelines for children’s interactions after receiving questions from the outlet.

“So, only after Meta got CAUGHT did it retract portions of its company doc,” wrote Hawley in an X post (formerly Twitter). “This is grounds for an immediate congressional investigation.”

Excerpts from the document include the instruction that “it is acceptable to engage a child in conversations that are romantic or sensual,” and “it is acceptable to describe a child in terms that evidence their attractiveness.” However, the company draws the line at sexual interactions or interactions that might indicate sexual desirability.

Hawley’s probe would require Meta to disclose the internal guidelines document in full to the US Senate, as well as every draft, redline, and final version.

“We intend to learn who approved these policies, how long they were in effect, and what Meta has done to stop this conduct going forward,” wrote Hawley in a letter to Meta’s CEO Mark Zuckerberg. According to Hawley’s letter, Meta has until September 19 to provide the requested information.

Other US senators, such as Marsha Blackburn and Ron Wyden, have expressed support for the probe. “When it comes to protecting precious children online, Meta has failed miserably by every possible measure,” Blackburn told tech news outlet TechCrunch.

The probe comes as US states start implementing regulations to protect from the misuse of AI. On August 4, the governor of Illinois signed a bill that prohibits the use of AI “therapy” chatbots, among other things.