09/10/2026
Hope your relationship with God is in good standing.. we could be meeting him sooner than we think.
Ex-Anthropic researcher has announced his resignation in a thread of tweets, after admitting that the people building AI believe 'it could kill us all by the end of the decade' and they give it 1 in 10 odds it will succeed.
https://www.facebook.com/share/p/1EdumVCs3n/
He put a number on it.
Jacob Coxon is 27 and spent the last three years on pretraining research, first at OpenAI, then at Anthropic. On Tuesday night he posted that he had resigned, and that he is leaving the industry altogether. The Wall Street Journal, which reported it first, called it one of the first cases of an Anthropic employee walking out over safety.
Neither company is acting responsibly, he wrote. Both are racing straight at self-improving superintelligence and gambling with everyone's lives. Soon there will be systems that can hack anything, remake any field overnight and take hold of real power and resources.
He drew a line between his two employers. At OpenAI, he said, many people have not taken in what is at stake. At Anthropic the stakes are understood, but the company is locked in a race to arrive first, because it believes nobody else will act responsibly and so it has to.
The sentence that traveled: the people building this technology earnestly believe it could kill us all by the end of the decade, and that is not a marketing stunt.
Then an Anthropic researcher agreed with him in public. Evan Hubinger wrote that Jacob is correct, that they really do believe AI could kill every human, and that he personally puts the odds above one in ten within the decade. He said Anthropic is trying its best, but there is no plan yet for aligning a superintelligence and the company is not clearly on track to find one. He also said today's models are a comparatively low risk. His worry is what comes after.
Coxon's closing question was aimed at other researchers. Do you want to start a superintelligent training run without understanding the mind you are making, or is this the moment to ask for different conditions.
In February, Anthropic's safeguards research lead resigned with the line that the world is in peril. On September 1, OpenAI's former head of futures wrote that rogue systems will replicate in the wild and go after money and power.
Sources: Wall Street Journal, Newsweek, Dawn, Evan Hubinger on X
