Post from Truth Social

A clip that repeats yesterday's post back to him

Twenty-four hours before this video went up, he had written: "The AI Hoax being perpetrated by the Radical Left Dumocrats is reminiscent of their Global Warming Scam of not so long ago." The segment he then published — a Fox Business exchange between Larry Kudlow and a guest identified as a "former White House deputy chief of staff" now running an "AI advocacy group" — makes the same move almost line for line: "they tried it as you pointed out correctly with the Paris Climate Accords it's not about the climate it's about control." The analogy he reached for on Sunday comes back to him on Monday night in a broadcaster's mouth, and he puts it on his feed without a syllable of his own.

That is the behavior worth reading. He is not adding an argument; he is collecting a confirmation and showing it to his audience as an independent one. The frame did not travel outward from him and get tested — it traveled outward and came home wearing a suit.

Praise received in the third person

Roughly a third of the transcript is direct laudation: "the greatest president for the worker, the greatest president defending America, America's national security," "he stood up and fought he stood up and led," "he sees around the corner on this," "he's not going to back down and kowtow." Publishing this wordlessly is a way of asserting it while keeping the fingerprints off. A self-authored "I am the greatest president for the worker" would be read as a boast; the same sentence sourced to a cable guest reads, to a sympathetic audience, as corroboration. The supply is real either way — what changes is deniability.

The superlative stack here is not incidental to the policy content. It arrives welded to it: the case against AI regulation is delivered as a case about his perceptiveness ("he sees through some of the the bull that we're seeing out there online from some of these companies"). Adopting the argument requires adopting the characterization.

The bad-faith premise, amplified

The clip's organizing claim — "this is not about safety it's about control," repeated three times and closed with "this is nothing to do about safety this is control in politics" — does not dispute that AI risk arguments are wrong. It asserts that no one making them believes them. That is an unfalsifiable attribution of motive, and it forecloses the debate rather than joining it. Applied to a named individual, it becomes devaluation: "take Dario the CEO of anthropic at his word he didn't say the American government he said the global world order," followed by "he wants to outsource the regulation the oversight to what the Europeans who are building nothing."

These are the broadcaster's words, not his, and they are not scored as his factual assertions. But the selection is an endorsement of the premise — and the premise is the same projection structure his own AI posts have been running: a stated concern about a risk is recoded as a concealed grab for power. The imputed motive is the one the frame's beneficiary is himself pursuing. His prior-day "Hoax" framing did the projection in his own hand; this does it by proxy, with a surrogate's credential attached.

The splitting is clean and total. On one side, "our two frontier labs are American companies built here in the United States of America," the president, the worker, national security. On the other, "the Democrats the left," "silicon elites," "the Europeans who are building nothing," and a global body invoked as "the global world order." Nothing occupies the middle; no AI safety claim in the segment is granted even provisional good faith.

What the clip stops short of

The hostility stays inside argument. Amodei is characterized as wrong and self-interested, not as subhuman or as someone to be dealt with physically; the Democrats are accused of wanting "control," not marked for removal. The electoral closing — "we're seven weeks out for an election" — routes the grievance to voting, which is where the prior post also sent it ("JUST GET OUT AND VOTE!!!"). There is no target-plus-action structure here, and reading one in would require importing it from elsewhere in the feed.

Authorship and cognition

With no written text, stylometry has nothing to work on. The timestamp converts to roughly 11:52 PM Eastern on September 14 — he had returned from Ireland by then, commenting publicly on Flock cameras that day — which puts it inside his long-established late-night window and matches his habit of pushing a TV segment hours after it airs. That evidence bears on who picked the clip, not on who composed anything, and it is thin enough on its own to hold the confidence low.

Nothing here supports a cognitive reading in either direction. He produced no sentences, so vocabulary, syntax and coherence are all unobservable; the transcript's rough edges ("we're split seven weeks out," "the global world order") belong to the speaker on screen and to the speech-to-text, not to him. Judgment of topical consistency is the one thing the post does show, and it is consistent: three consecutive days on AI regulation, the same analogy, the same imputed motive.