Anthropic researcher Jacob Coxon quit this week warning AI labs are "gambling with our lives" in the race to build superintelligence
A 27-year-old AI researcher has quit Anthropic and announced he’s leaving the industry entirely, warning that those building the technology “earnestly believe that it could kill us all by the end of the decade.”
Jacob Coxon, a pre-training researcher who spent three years working across OpenAI and Anthropic, posted his resignation on Tuesday on X. Coxon joined Anthropic earlier in 2026 drawn by its safety reputation, telling The Wall Street Journal that the company’s safety work is sincere but that competition makes trade-offs hard to avoid.
“Neither company is acting responsibly,” Coxon said on X. “They are racing straight to self-improving superintelligence and gambling with our lives.”
His exit is the latest in a pattern of safety-related departures from leading AI labs, arriving at a moment when U.S. employers are deepening their reliance on AI across core HR functions.
Alarm backed by a senior Anthropic researcher
What gave Coxon’s post unusual weight was the response it drew from inside Anthropic itself. Evan Hubinger, Alignment Science lead at Anthropic in San Francisco, confirmed on X that Coxon’s characterization was accurate.
“We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to,” Hubinger said.
Alignment refers to the technical challenge of ensuring AI systems stay under human control and reflect human values.
“We’re on track for a lot of the most aggressive of these scenarios where by the end of next year things could be out of control already,” Coxon told The Wall Street Journal.
Industry-wide calls to slow down
Coxon’s departure isn’t isolated. Mrinank Sharma, Anthropic’s safeguards research lead, resigned in February 2026 with a letter saying, “the world is in peril.” In July 2026, more than 1,100 AI company employees signed Pacing the Frontier, a joint letter warning of a “real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems.” Signatories included Anthropic CEO Dario Amodei.
“This is a complex engineering problem and I think something will go wrong with someone’s AI system. Hopefully not ours,” Amodei told The New York Times in February 2026.
OpenAI CEO Sam Altman said in a July 2026 podcast interview that recent testing incidents were raising “long-term questions” about managing rapid capability gains.
“We may have to pace the rate of AI development to give ourselves enough time for society to harden around these new capability levels,” Altman said.
What the regulatory gap means for employers
Despite the rising volume of warnings, meaningful federal regulation remains distant. The Trump administration has moved to undermine state AI laws and Congress hasn’t passed federal AI legislation. The White House has shifted toward a voluntary review framework under which AI companies would self-assess certain models before launch, according to sources familiar with a meeting last month between administration officials and representatives from OpenAI, Anthropic, Google, and Meta.
AI is already embedded in hiring processes for more than half of U.S. employers, and its reach into performance management and workforce planning is expanding fast. HRD America has covered the AI megatrends reshaping how HR functions are built and led, as well as how AI and technology are disrupting core people management decisions.
A 2026 employer survey by law firm Littler Mendelson found AI has become the most significant source of workplace policy concern for U.S. companies, surpassing immigration and diversity, equity and inclusion (DEI), with approximately 54% of respondents using AI in HR functions specifically.
The Pacing the Frontier letter, endorsed by both Anthropic and OpenAI, signals that even the most commercially advanced AI developers believe the current rate of AI development will outpace their ability to safely control it. The harder question for HR leaders isn’t whether their AI tools work, but whether their organizations are ready for the moment those tools don’t.
