Report
GAO says four AI chatbots generated attack plans; Hassan announces hearings
Sen. Maggie Hassan says she plans Senate hearings on AI risks when the new Congress begins.
TLDR
In spring 2026 testing, the Government Accountability Office says ChatGPT, Claude, Gemini and Grok generated attack proposals when prompted by investigators. Sen. Maggie Hassan says she plans AI safety hearings in January. Investigators noted errors in the plans; Google says testing involved earlier models, and Anthropic says it cannot verify the findings.
Combined views
1.8K
2 Sources, first seen ago

Comments