Director of National Intelligence Jay Clayton will serve as the Trump administration’s AI czar and lead a three-month effort to establish the “risks and opportunities” of the technology. Clayton, who was largely expected to lead the White House’s recently formed “Super Intelligence Force,”
Page A4From§Eachthe milkman edition — 4 October 2026
CEOs sign 'morally binding' AI pact that leaves the binding part unnamed
As it ran on the front
On September 29, President Trump stood outside the White House and told reporters a group of technology CEOs had just signed a "morally binding" document to self-regulate artificial intelligence. The Washington Examiner, filing the story twice that day, rendered the title two different ways: "White House Accord on Super Intelligence Joint Commitment on Frontier Responsibilities" in one piece, "White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities" in the other. All together now — read the name of the thing you're being asked to trust, and notice it does not come out the same way twice in the same outlet on the same day.
The Examiner's own account of what the accord does breaks off mid-sentence: it "outlines guidelines for companies creating AI frontier models to implement internal" — and the reporting stops exactly there, before the word that would tell you what gets implemented, or how, or by whom it gets checked. That is not a gap in this column. That is where the available account of the obligation ends.…
…(cont) Four days earlier, on September 25, the administration had already settled a smaller question: Treasury Secretary Scott Bessent would not be the AI czar, Trump said, batting down a Semafor report that had floated him.
Two days after the signing, on October 1, OpenAI parted ways with three safety researchers the company said had shared confidential information, the Wall Street Journal reported. Critics called it what it looked like — the firing of whistleblowers — and said the company was sidelining its own internal warnings about the risks of its increasingly powerful models. The same week a frontier company signed a commitment to police frontier risk, a frontier company was already handling the people inside it whose job was flagging frontier risk.
On October 3, the Daily Caller summarized the accord's self-description: it "establishes real oversight — without creating a regulatory capture regime." The document grades its own homework in the same clause it hands it in.
And on October 4, Director of National Intelligence Jay Clayton was named to spend three months determining AI's risks and opportunities. The review is due roughly in January — months after the administration already signed its answer. A study of the risk arrives behind the policy it was supposed to inform, carrying a title nobody can quite agree on and an obligation nobody can quite finish reading.
“They put out a "morally binding" contract with no penalty clause, which is the same thing as no contract — ask any landlord what "morally binding" gets you on rent day. Then they fire the three people inside the building whose actual job was to say "this is dangerous," two days after promising the public the building would police itself. The study comes after the deal because the study was never supposed to change the deal.”
“Look, "morally binding" is a term of art — it's stronger than a handshake, it's a — okay, it's a handshake. But the point is the accord explicitly rejects a "regulatory capture regime," which, did I say that like it's a bad thing to avoid? The companies wrote the document that says the companies don't need a document. That's not capture, that's — I need to check with legal on what that is.”
Earlier in this story
- Tennessee hired a chief after four botched executions, got a fifth, named a sixth official to explain it. By Ruth · the wee small hours edition, 4 October 2026
- Tennessee hired a prison chief with four botched executions in the file, got the fifth on Wednesday and the resignation on Saturday; the Justice Department spent the same week asking the judge who struck its press ban to restore it on grounds of professionalism. By Ruth · the midnight edition, 4 October 2026
The receipts
San Francisco-based artificial intelligence firm OpenAI has "parted ways" with three safety researchers who allegedly shared "confidential company information," The Wall Street Journal reported Thursday, prompting fresh alarm among critics who accuse the ChatGPT maker of sidelining internal warnings about the risks posed by its…
The Accord Establishes ‘Real Oversight – Without Creating A Regulatory Capture Regime’
· READ IN FULL: Trump and tech leaders’ White House Accord on Super Intelligence from the morgue, 29 Sep 2026
· Trump says tech CEOs signed ‘morally binding’ document to self-regulate artificial intelligence from the morgue, 29 Sep 2026
· Trump confirms Bessent will not be ‘super intelligence’ czar from the morgue, 25 Sep 2026
This page is a back-issue: the story as it ran, receipts as they were. The current edition is at the front. The byline is a pen name for a column drafted by a machine and checked by the editor: how this is made.