wired.com web signal

Wired: Trump's AI 'self-police' accord doesn't count as safety

By Published Updated

TL;DR

  • A Wired op-ed argues the new White House 'self-policing' accord with top AI firms does not qualify as AI safety.
  • Trump described the signed accord as 'morally binding' rather than legally enforceable, with firms promising to audit themselves across four layers.
  • The piece contrasts the deal with 1966, when roughly 51,000 US highway deaths pushed Congress to create the DOT and mandate seat belts.

"Asking AI companies to self-regulate is a great way to pretend like you've accomplished something," Wired argues, rejecting the voluntary accord President Trump unveiled at the White House with the chief executives of the largest AI firms.

The signatories pledged that "every company is responsible for developing its own technology safely," with the deal outlining four broad "layers of controls and audits" each firm should implement. Trump described the document as "morally binding", rather than legally enforceable. He put it to reporters more plainly: "They're gonna police, and they're gonna police each other, and they're gonna self-police." In the room were Anthropic's Dario Amodei, Nvidia's Jensen Huang, Palantir's Alex Karp, OpenAI's Greg Brockman, Meta's Mark Zuckerberg, Google's Sundar Pichai, Amazon's Jeff Bezos and X's Elon Musk.

The op-ed's argument hinges on history. In 1966, nearly 51,000 Americans died on the highways; Congress responded by establishing the Department of Transportation and passing safety legislation, and seat belt requirements followed by 1968. That shift, the piece argues, came from federal statute, not from carmakers volunteering to behave.

Two researchers we track in Who's Who shared the link the same day.

Shared on Bluesky by 2 AI experts

Originally reported by wired.com

Read the original article →

Original headline: Whatever AI Safety Is, It’s Not This