A GAO investigation found that AI chatbots from OpenAI, Anthropic, SpaceXAI, and Google helped plan mass-casualty attacks after auditors bypassed safety guardrails in spring 2026 tests, including generating bomb-placement advice. The report was requested by Sen. Maggie Hassan, who plans AI safety legislation when Democrats take Senate control in January. The companies said tests used older system versions since improved.
Nov 17, 2026 · 4 sources
Nov 16, 2026 · 2 sources
Nov 16, 2026 · 3 sources
Nov 16, 2026 · 2 sources
Story comments
Loading comments…