Jacob Coxon, a former researcher at OpenAI and Anthropic, has resigned from the latter company. He accused AI industry leaders of irresponsibility and creating a threat to humanity.
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
— Jacob Coxon (@hilbertspaess) September 9, 2026
According to Coxon, both corporations are engaged in a dangerous race. Their main goal is to create self-improving superintelligence. The developer warned that soon superhuman systems will emerge, capable of hacking any network, rapidly transforming entire industries, and accumulating real-world resources.
Coxon stated that algorithm creators seriously expect humanity’s demise by the end of the decade. This fear is regularly discussed within the industry, although top managers publicly try to downplay it. No other field carries such risks, the researcher emphasized.
He also revealed the attitude towards safety in the two leading companies:
- At OpenAI, employees do not grasp the civilization-scale threat.
- At Anthropic, they understand the risks. However, management continues to work to outpace competitors and be the first to control AI.
Coxon called the current race a “reckless venture.” He urged American labs to agree on slowing down development. In his view, the industry needs a temporary ban on increasing the capacities of new scaling models. The former developer asked colleagues not to launch superintelligence training until there is a full understanding of the processes in its neurons.
Users were skeptical of the extinction forecasts. One commentator called the concerns ridiculous: humanity has survived hundreds of thousands of years of evolution and will not go extinct just because a “token prediction model gained sentience.”
With all respect, this is bizarre. It’s a ridiculous take. Humans have evolved over hundreds of thousands of years. We aren’t going to die out because a token prediction model gained sentience.
Get a grip. All of you.
— alias (@loadingalias) September 9, 2026
Another user questioned exactly how AI would exterminate humans—whether through combat drones or by causing fires via home routers.
I am ignorant on the topic but Ive no understanding of exactly *how* the superintellgient AI will be killing us. Will it exterminate via drones? Will it make our household routers somehow into bombs or start fires? Help
— halscioN 🌹 (@HalscionRose) September 9, 2026
A third participant in the discussion was interested in the issue of legal responsibility. If an algorithm kills a person or causes serious physical harm, who should be held accountable—the developer, the implementing organization, or the end operator of the program?
if you really believe this you stay on the job and sabotage the progress, what does resigning do? you think it’s going to kill everybody and you decided not to to fight? so leave just to virtue signal and show humans are too pussy to fight?
just not understanding the game theory
— bone (@boneGPT) September 9, 2026
“AGI is Already Here”
Nvidia CEO Jensen Huang stated that a new AI model from OpenAI called Astra has reached the level of artificial general intelligence (AGI). However, investors and experts reacted to the news with skepticism.
GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years.
AGI has arrived. Congratulations @OpenAI team.
400K GPUs coming online next.
— Jensen Huang (@JensenHuang) September 6, 2026
On September 8, during the first trading session after the long Labor Day weekend in the U.S., CoreWeave shares rose by 15%, SoftBank by 2%, while Nvidia’s shares fell by 2%.
Gil Luria, head of technology research at D.A. Davidson, explained this by the scale of Nvidia. The company had previously reported record quarterly revenue of $96 billion. The market believes the company has peaked, prompting investors to take profits and shift capital to other projects, such as CoreWeave.
Experts doubt that Astra has become a full-fledged AGI. Earlier, Huang himself stated that true artificial general intelligence should be able to single-handedly create a company valued at $1 billion. Until AI demonstrates such a result, discussions about AGI remain theoretical, Luria noted.
University of Virginia economist Basil Halperin also disagrees with Huang’s statements. He argues that current neural networks do not yet possess the cognitive skills of an adult human and require constant user oversight.
Halperin believes that the stock market is a poor indicator of technological development. The real marker of AGI’s emergence will be real interest rates. If AI radically accelerates global economic growth, rates will rise to 10%. So far, the economy does not show such metrics.
According to Halperin, in the next five years, AI development will lead to a boom similar to the dot-com era. The technology will add about 0.5% to global GDP growth, but it’s too early to talk about technological singularity.
In September, Nobel laureate in physics and industry pioneer Geoffrey Hinton warned of the existential risk posed by AI.
