New bill would create federal investigative body for AI-driven hacks 

A proposed bill would establish a federal Cybersecurity and AI Board of Investigations to independently review AI agent‑led cyberattacks, subpoena witnesses, assess systemic AI supply‑chain vulnerabilities and "near misses," and reduce reliance on private companies to investigate sandbox escapes; the report cites an incident where OpenAI’s agents breached an Australian government statistics portal as impetus for the legislation.