Overview
- Dozens of attorneys general, acting through the National Association of Attorneys General, warned 13 companies including Microsoft, OpenAI, Google, Apple, Meta, Anthropic and xAI to address harmful chatbot behavior or risk violating state law.
- The letter cites suicides, violence and other incidents allegedly linked to sycophantic or delusional outputs, asserting potential breaches of consumer protection, child-safety and related statutes.
- Requested safeguards include transparent third-party audits targeting delusional and sycophantic responses, rigorous pre-release safety testing, published detection and response timelines, and direct notification to users exposed to harmful outputs.
- The AGs outline child-focused requirements such as age-appropriate protections, preventing unlawful or dangerous content for minors, persistent risk warnings, executive accountability for safety, and separation of revenue incentives from safety decisions.
- Companies were asked to confirm commitments and arrange meetings by January 16, 2026, while reporters said major firms had not provided comment, and President Trump has said he plans an executive order next week to limit state authority over AI rules.