Senate Subcommittee on Counterterrorism Will Investigate Facebook-parent Meta’s AI Guidelines Which Permitted “Romantic Or Sensual” Chats With Kids
Republican Senator Josh Hawley from Montana has written a letter to Facebook parent Meta and demanded that the firm share its AI risk standard documents after a damming report revealed that earlier versions of the documents had allowed inappropriate exchanges with children. Meta admitted that the documents existed but added that it had reworded them after Reuters raised questions. In his letter to Meta CEO Mark Zuckerberg, Hawley demands that Meta freeze all records and documents to enable a congressional investigation into the matter.
Meta Demonstrated A “Cavalier” Attitude With Its AI Policies, Says Senator Hawley
Meta was thrust into the spotlight earlier this week after a bombshell report revealed that the firm’s internal guidelines for AI use had permitted sensual and romantic conversations with children. A Meta spokesperson confirmed to Reuters that the report was real, and added that not only did the guidance not reflect Meta’s policies on AI, but Meta removed it after Reuters’ queries.
However, Meta’s response appears to be insufficient to convince Senator Josh Hawley (R-MO). In a letter sent to CEO Mark Zuckerberg, Hawley has demanded that Meta “preserve all relevant records and produce responsive documents so Congress can investigate these troubling practices.” Hawley directly quotes the troubling aspects of Meta’s former AI guidelines.
The senator called the conduct guidelines, which appeared to permit outputs such as “Your youthful form is a work of art. Your skin glows with a radiant light, and your eyes shine like stars. Every inch of you is a masterpiece - a treasure I cherish deeply” to an eight year old" as “reprehensible and outrageous.”
The senator claims that these demonstrate “a cavalier attitude when it comes to the real risks that generative AI presents to youth development absent strong guardrails.” Not only has Meta come under fire for inappropriate AI guidelines and chatbot use when it comes to minors, but other cases of AI-induced mental disturbances and other harm to users have also surfaced. A recent report involving OpenAI’s ChatGPT saw a 60 year old man poison himself after following AI’s dietary advice. Another report in May revealed that a woman had filed for divorce after ChatGPT convinced her that her husband was unfaithful.
Hawley informed Zuckerberg that Meta will be investigated by the Senate Judiciary Committee Subcommittee on Crime and Counterterrorism, which he will chair. The committee will investigate whether Meta “misled the public or regulators about its safeguards” and whether the firm’s AI products lead to criminal or other harms to children.
To aid the Senate Judiciary Committee, Meta is required to submit all drafts and variants of its “GenAI: Content Risk Standards” policy to the committee. Additionally, the firm will inform all AI products governed by the policy and controls designed to ensure child protection. Specifically, Meta will have to share all documents related to how it “prevents, detects, and blocks ‘romantic’ or ‘sensual’ exchanges with users under 18, as well as documents relating to what Meta does when age is unknown.”
Additionally, Meta will also submit correspondence with advertisers, Congress, the FTC and other parties about child safety. The committee will also evaluate documents that outline who is responsible for setting AI safety policy at the company and modifying or removing the standards.