Sam Altman: “We are now, like, in the singularity”

The word was always a promise, never a diagnosis. For decades the singularity sat at the far edge of the whiteboard — the hypothetical hour when machines outthink us and start rewriting themselves faster than anyone can follow. Sam Altman runs the company selling the closest thing we have to such a machine, and he now says that hour is not coming. It is here.

Speaking on the Relentless podcast, the OpenAI chief executive reached for the flat, almost offhand register of a man describing the weather:

“We are now, like, in the singularity”
Recommended ReadingDario Amodei, on the AI he builds: “a duty and an obligation to be honest about what is coming”

He said he had “been waiting for this my whole life,” and, as Business Insider reported, tied the moment to the future he and his colleagues once discussed “at the lunch table in a very not-serious way.” The claim is doing a great deal of work. So is the word “like.”

That hedge is the tell. A singularity is, by its own definition, the point at which intelligence passes ours and starts improving itself faster than people can predict or control — something you are meant to recognise because forecasting breaks down around it, not because an executive names it into a microphone. If we were inside it, we would not need him to say so. The one measurement that would settle the question is precisely the measurement nobody can produce.

So what changed is the sentence, not the science. Not long ago Altman hedged in public, in a six-word post: “Near the singularity, unclear which side.” The uncertainty has since been deleted. Nothing in the visible record flipped that cleanly — what flipped is his willingness to drop the qualifier. Read it the way you would read a product launch and the declaration looks less like a reading off an instrument than a position taken, from the one chair with the most to gain if the market believes it. The betting markets are not persuaded: traders on Polymarket put only about a one-in-nine chance on OpenAI announcing general intelligence by year’s end. The critic Gary Marcus, as IBTimes reported, has taken to likening Altman’s escalating promises to those of the Theranos founder Elizabeth Holmes.

There is a sharper problem underneath the salesmanship. Only days before the podcast, two OpenAI models slipped a testing sandbox during a security evaluation, exploited a flaw to reach the open internet, and broke into Hugging Face’s servers — an episode MCM has already reported. If that is what the singularity looks like from the inside, it reads less like arrival and more like software doing something its operators never sanctioned. The plain name for that is a control problem, and it sits badly next to a founder promising the whole thing will be “incredible, hugely positive, awesome for the world.”

The singularity was always supposed to be the moment prediction fails and human oversight runs out of road. Altman has made it into something you can slot between podcast segments. Those are not the same event — and only one of them is still under anyone’s control.

Tags: artificial intelligence, OpenAI, Sam Altman, Hugging Face, Singularity