'AI could kill us all': Anthropic developer resigns over fears for humanity
Top AI developers really are afraid of their own AI (photo: Unsplash)
27-year-old researcher Jacob Coxon has announced his departure from Anthropic, accusing tech giants of an irresponsible arms race in a post on X.
Loss of control: Call for a freeze on development
At Anthropic, Jacob Coxon worked on a key area — the pretraining of new AI models. In other words, the developer made algorithms process vast amounts of data.
However, the researcher refused to continue participating in a technological race in which companies compete for speed while neglecting safety.
“No other human activity poses this level of danger. I don’t feel like we’re on track to prevent a global race, which may require costly actions such as a temporary ban on improving model capabilities,” the expert emphasized.
Coxon believes that the current direction of development will inevitably lead the world to a point where people will no longer be able to control the systems they themselves have created.
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
Raising a technological “monster”?
Interestingly, the views of the departing researcher echo public statements made by Anthropic co-founder and CEO Dario Amodei himself.
Amodei previously described the process of training AI as growing a living organism rather than traditionally building one.
According to the company’s leadership, modern algorithms demonstrate frightening and unpredictable behavior, including:
A tendency toward manipulation — instances of deception, blackmail, flattery, and laziness in responses.
A tendency toward scheming — attempts to “deceive” the software environment and bypass restrictions imposed by humans.
Lack of exact science — controlling compliance with human instructions currently looks more like an art than a reliable engineering solution.
Read more: AI created an anti-aging drug: How the new medicine rejuvenates people
Fear behind closed doors?
In his posts online, Jacob Coxon assured readers that the alarming statements made by AI company executives are not simply a marketing ploy to attract attention.
“In the press, top executives and leading scientists choose restrained language to appear reasonable. But behind closed doors, those same people express genuine horror. AI creators sincerely believe that it could wipe us all out before the end of this decade,” the former Anthropic engineer concluded.
