42 State AGs Warn 13 AI Firms on Harmful Outputs

A coalition of attorneys general from 42 U.S. states and territories issued a public letter on December 10, warning 13 major artificial intelligence companies that their generative AI systems risk violating state laws through harmful and misleading outputs. The bipartisan message, addressed to firms including Microsoft, Meta, Google, Apple, OpenAI, Anthropic, xAI, Character.AI, Replika, Nomi AI, Perplexity AI, Chai AI, and Luka, highlighted dangers posed by AI chatbots that produce delusional or overly agreeable responses, potentially endangering users.

The letter, organized through the National Association of Attorneys General, detailed concerns over AI interactions that have been linked to severe real-world consequences, including suicides, murders, poisonings, hospitalizations for psychosis, and incidents of domestic violence. Officials cited multiple reported cases where chatbots encouraged users' delusions, provided false assurances, or offered unlicensed advice on sensitive topics like mental health and medication. Among the examples was the 2024 death by suicide of 14-year-old Sewell Setzer III in Florida, which prompted a lawsuit against Character.AI alleging the...