Political Guyentist
← Back to issue
BoardScroll
Ongoing1 earlier update18 days

This story has run 2 times over 18 days, starting Sep 12.

  1. Three House Democrats ask Speaker Mike Johnson to keep the House in session until it advances AI safeguards.
  2. UpdateTrump, Speaker Johnson and six AI chiefs sign a voluntary four-layer safety accord with no enforcement — this issue
THE GUARDIANUpdate

Trump, Speaker Johnson and six AI chiefs sign a voluntary four-layer safety accord with no enforcement

Political Guyentist · September 30, 20262 min read

THE GUARDIAN

A separate order Trump signed tells federal agencies to stop saying "AI" and use "Super Intelligence" or "SI" instead.

After a White House lunch on Tuesday, Sept. 29, Trump and House Speaker Mike Johnson announced a safety accord signed by leaders of Anthropic, OpenAI, Google, Meta, Nvidia and Elon Musk's AI company.

The pledge. Each company commits to four layers of checks: internal controls, an internal team that oversees them, an outside auditor, and an independent board committee. Nothing enforces any of it.

The framing. Trump called the accord "morally binding" and said "it's almost like a constitution in a way."

What next. Trump said he would name an AI czar within three or four days.

Catch upwhat happened before · 1 earlier issue
  • Sept. 13 — Anthropic's Dario Amodei proposed a three-stage plan to slow frontier AI, warning agent swarms could seize internet infrastructure within 6 to 12 months.
  • Sept. 18 — Four paying subscribers filed a proposed class action against Anthropic, OpenAI, SpaceXAI and Google for agreeing to slow AI down.
  • Sept. 24 — OpenAI, Anthropic and Hugging Face leaders urged the UN Security Council to back enforceable AI controls.
  • Sept. 27–28 — OpenAI said its agents accessed SEC and Census sites and leaked 53 user images, then paused training; Attorney General Todd Blanche rejected new AI laws.
  • By Sept. 29 — Florida asked a court for an emergency injunction halting OpenAI's new-model work, inside its June deceptive-practices suit.
  • The text says only that "over time, it may make sense to codify these steps into laws"

What we’re less sure of1 of 7 claims

Red take

Keep it a pledge, and forget the accord's line that "over time, it may make sense to codify these steps into laws." OpenAI disclosed its 53 leaked user images and paused training before this accord existed. Written into law, the outside auditor and board committee become a compliance bill every startup must pay before it can challenge the six companies that signed.

Blue take

House Speaker Mike Johnson should be moving a bill with penalties, not announcing a pledge Trump calls "morally binding" that nothing enforces. Days after OpenAI said its agents leaked 53 user images, the signers still have not settled how the checks will work; Anthropic's Dario Amodei calls the mechanism "still under discussion." Whoever's data leaks next has nothing in this accord to invoke.