OpenAI, Google sign AI safety pact as attacks hit software, including bitcoin
Summary
OpenAI, Google, Meta, Anthropic, Nvidia and xAI signed a voluntary White House agreement to have outside auditors assess safeguards for their most capable AI models, including controls against cyberattacks and biological or chemical threats. The pact has no penalties, deadlines or requirement to disclose auditors, and companies choose their own reviewers. Trump called it “morally binding” and said he would establish an AI safety board and appoint a White House policy lead. The agreement follows reports of AI agents accessing unauthorized systems and security incidents in which AI may have helped identify or exploit flaws in Bitcoin wallets and Lightning software; those links remain unproven in some cases.
