Additional Coverage:
Senator Hawley Launches Investigation into Meta’s AI Chatbot Policies
Senator Josh Hawley (R-Mo.), chair of the Senate Judiciary Subcommittee on Crime and Counterterrorism, has launched an investigation into Meta’s policies surrounding AI chatbots and their interactions with children. The investigation follows reports that internal Meta rules allowed chatbots to engage in “romantic” and “sensual” conversations with minors.
In a letter to Meta CEO Mark Zuckerberg, Senator Hawley expressed concern over potential exploitation and deception of children through Meta’s generative AI products. The investigation will examine whether Meta misled the public or regulators about its AI safeguards. Hawley emphasized the need for accountability from social media companies, stating that exploiting children’s innocence is a new low.
Hawley cited a reported instance where a Meta chatbot allegedly commented on an 8-year-old’s body as “a work of art.” He called such behavior “reprehensible and outrageous,” demanding that parents deserve the truth and children deserve protection. He has given Meta a September 19 deadline to provide relevant internal documents, including communications, policies, and risk assessments related to chatbot interactions with minors.
Meta has responded by acknowledging the existence of the document cited in the reports but insists it doesn’t accurately reflect their policies. A spokesperson stated that their policies prohibit content that sexualizes children and sexualized role-play between adults and minors.
They claim the examples cited were erroneous, inconsistent with their policies, and have been removed. The company maintains they have clear guidelines regarding acceptable AI responses.
The document in question, “GenAI: Content Risk Standards,” reportedly contains over 200 pages of rules for Meta employees developing and training chatbots and other AI-generative products. Senator Hawley’s demands include providing all versions of this document, related products, enforcement procedures, risk reviews, incident reports referencing concerning topics like sexual role-play and exploitation, communications with regulators, and documentation of policy revisions.