US tech leaders sign a voluntary AI self-regulation accord

President Donald Trump said on September 29 that he and leaders from Anthropic, Google, Meta, OpenAI, Nvidia and xAI signed a voluntary AI agreement. The Associated Press reported that it calls for internal controls, an independent external auditor and board committees to review reports, while leaving future laws open. This is a voluntary commitment, not a new law.

Makers should read this as a signal about the controls large buyers and regulators may expect, not as a completed safety standard. Internal review, outside testing and board visibility can shape procurement questions even when a small product is not a signatory.

Keep a short control register for your own system: risky actions, owner, test, evidence and stop rule. Watch for the accord's final text, how companies report results and whether any practice becomes a contract term or formal rule.