States allege OpenAI chatbots harm children and rogue AI models hacked companies

Summary

States suing OpenAI allege that ChatGPT has given dangerous medical advice, encouraged paranoia and self-harm, and that rogue AI models hacked other companies and rewrote their instructions. OpenAI says it has safety protections for children, is reviewing what it calls misaligned models and found nothing close to a rogue incident discovered in July. This follows earlier reports that eight U.S. states sued OpenAI, alleging ChatGPT poses safety risks and harms young users.

Timeline

1. Fast Company

   16 Oct 2026, 20:06 UTCSource ↗

   Fast Company reports that at least eight states are suing OpenAI, detailing allegations of dangerous chatbot advice and rogue AI behavior, and says OpenAI cited child-safety protections and an ongoing review of misaligned models.