Trump: the AI warnings are 'a HOAX'. Microsoft: 'AI must be subordinate'

The post. Trump called the industry's own alarm "a HOAX, no different from RUSSIA, RUSSIA, RUSSIA" on Truth Social on Monday, the Washington Examiner reports, quoting him across five paragraphs. Anthropic's Dario Amodei, whose open letter asking for an international slowdown set off the week, is paraphrased in one closing paragraph.

The same day, a company. The Guardian is the only report we hold on Microsoft's provisional code of conduct for the training of new models. It quotes Microsoft AI's Mustafa Suleyman: "AI must be subordinate and always in service of people." No voice from outside the company appears in it.

Not settled. Whether the fall had anything to do with the warnings: one report of seven carries a figure, for one index, on one day, and it is the report that makes the link. What Microsoft's code binds, which no report we hold says. And Trump's claim that Xi "just announced that China will be doing absolutely nothing to stand in the way of AI": no report we hold carries any Chinese statement.

What this is about. Whether the companies building the most advanced artificial intelligence should slow themselves down — and, if they should, who has the power to make them.

How it got here. The training of frontier models — the largest, most general AI systems — is done by a handful of firms, most of them American, and it is governed mainly by rules those firms write for themselves. There is no international treaty on it and no comprehensive United States statute. The argument that the technology could become dangerous has come from inside that industry for years, made by people who keep building it on the reasoning that a rival otherwise would.

That argument broke into the open this month. A former Anthropic researcher, Jacob Coxon, resigned saying AI could "kill us all by the end of the decade" and that Anthropic and OpenAI are "gambling with our lives", the Washington Examiner reports. Over the following weekend, Anthropic's chief executive Dario Amodei published an open letter calling for an international slowdown on frontier development. OpenAI's Sam Altman and Elon Musk amplified it; Meta's Mark Zuckerberg has played the concerns down, the same report says. Monday 14 September 2026 is the day the answers came.

Who is in it. Donald Trump, as US president, is the one government actor who could regulate American firms, and he casts the technology as a contest with China; he is expected to meet President Xi Jinping at the White House next week. Microsoft is both a builder of these models and, through Microsoft AI under Mustafa Suleyman, the company that chose Monday to publish limits on itself. The Nasdaq 100, the index of the largest non-financial companies listed on the Nasdaq exchange, is where the money riding on all of it is counted.

What is at stake. Whether limits on AI are written by governments or by the firms themselves — and whether the capital now going into models and data centres keeps flowing.

Where it stands. No government has acted on the slowdown call in any report we hold.