Link: https://jacobin.com/2026/09/openai-anthropic-human-extinction-criti-hype

The models did what they were instructed to do. Nothing went rogue. As Cal Newport writes, “Circumventing internet restrictions and hacking into servers are exactly the kinds of things these ExploitGym systems are designed to do.” If the example reveals anything, it is how sloppy OpenAI was in controlling ExploitGym’s parameters for the sake of competing with Anthropic. And it cannot have been unhappy with the resulting coverage that, to the extent that it was alarmist, was also a tribute to the power of its models and to the ultimate objective of creating an AGI.

The obverse of criti-hype is just hype. A recent example of this was OpenAI’s claim to have solved the Navier–Stokes Millennium Prize problem. In essence, the problem is whether the standard mathematical equations used to describe the motion of fluids (the Navier–Stokes equations) are mathematically perfect under all physical conditions. OpenAI said it had used a swarm of ten thousand AI agents over eighty-eight hours to solve the problem definitively, proving that the equations always reach a limit: since “a real fluid cannot move infinitely fast, this would mark a breakdown in how the equations model the fluid.”

OpenAI’s claim was totally disingenuous: it had gotten wind that other researchers, Tristan Buckmaster and Levent Alpöge (who works at Anthropic), had made critical advances to the solution and used their pathways to prompt its own agents. This is really an appropriation of the commons, in this case the mathematical commons — the whole process of solving the problem from Claude-Louis Navier and George Gabriel Stokes onward has been a collaborative human effort.

OpenAI’s solution is not yet peer reviewed, and the Clay Mathematics Institute, which runs the Millennium Prize, lists the problem as unsolved. Even so, like the Hugging Face hack, this entered into internet lore as an example of how rapidly AI is accelerating toward AGI and out of human control, despite the fact that not one of these AI models has ever approached such a threshold, nor done anything that it was not designed, enabled, and instructed to do.

It's true that tech companies have an incentive to lean into a narrative about the risk of a rogue superintelligence and publish misleading exaggerations about how imminent that risk is, but that isn't the same thing as having a strong case that no such risk exists. The article's case for this seems pretty strange to me:

Obviously, we must plaster a huge advisory label on talk of “goals.” Unlike an organism, a large language model (LLM) doesn’t have internal goals any more than a thermostat or a motion detector does. An LLM is just a fixed function that iterates whenever it is used. It has no interiority, no subjectivity, no active persistence over time; any goals it did have would be the human purposes reflected passively in its functional design. At best, ascribing goals to the AI itself is a case of researchers adopting the “intentional stance” because it predicts well in a range of tested circumstances. We can thus, with economy, speak of the “loss function” used in the training of large language models wherein they are programmed to bring the gap between their prediction and the correct answer as close to zero as possible, as a “goal.”

In that case, however, what we are describing as a goal is just an observed regular output of a certain engineering; it is not a real cause. In the Bostrom sense, however, these ascribed goals somehow become real causes, capable of generating new goals.

For the purposes of whether something is apocalyptically dangerous, what it is actually able to do seems like it should matter a lot more than whether the same words used for humans can apply philosophically.

Since large language models can’t “hack” anything, merely plausibly continue a tokenized sequence, they were combined with a coding harness — a software infrastructure layer that allows a large language model to act as an autonomous coding agent — that was specifically optimized for completing cyberexploitation tasks.

A LLM with a coding harness can write and modify a coding harness. It's fair to point to the huggingface hack example as an example of OpenAI giving an exaggerated impression of how dangerous and capable their product is, but you can't pretend that the capabilities of a LLM based agent as a whole system are not the real goalposts here. A LLM plus its harness being able to produce another LLM plus its harness that will do very dangerous and consequential things, that's what such a risk could look like. The question is how good it is at building on itself without human assistance, and whether there are fundamental limitations that may prevent it from ever getting much better at it. Whether such limitations exist and what they are is not a settled issue, and so civilization level risk can't rationally be taken out of consideration.

The earth system is collapsing from climate change, I don't need to indulge your scifi speculation about nonsense hypothetical existential threats.

That is true, you don't need to. Articles like this one can't pretend they have seriously considered it though, just like all the dismissive rhetoric over the decades about climate change was more of an attempt to propagate an opinion than it was a serious consideration of what is actually going to happen to the planet given what we know.