Current and former employees of Anthropic, an American AI company founded by former OpenAI leaders, say they believe AI could potentially kill all humans within the next 10 years.
“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,” Anthropic alignment-science lead Evan Hubinger wrote Tuesday night on X.
His comments came after Anthropic AI researcher Jacob Coxon resigned from the company and took to X to detail his departure.
“The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon stated. “This is not a marketing stunt. If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible – but I hear the same people express fear privately. No other human activity poses this level of danger.”
The remarks come as AI becomes an increasingly routine part of daily life for millions of Americans, even as fears over the technology’s potential dangers intensify. Concerns about AI becoming uncontrollable are not new, as Elon Musk and leading researchers and academics have repeatedly warned that increasingly powerful systems could pose serious risks to humanity.
open image in galleryAround 2 a.m. Wednesday, Anthropic scalable-oversight lead Samuel Marks also responded to Coxon’s post, claiming that he was “[Writing this in a personal capacity, not on behalf of my employer (Anthropic).”
“AI developers believe their technology could cause human extinction (or similarly bad outcomes). This could happen in the next few years. In general, the more senior the employee, the more concerned they are,” Marks wrote.
The Independent has contacted Anthropic and Coxon for comment.
OpenAI, a separate but leading AI company, declined to comment on Coxon’s post when contacted by The Independent. However, a spokesperson pointed to several updates the company has published in recent weeks outlining its efforts to improve AI safety and alignment. This includes research into monitoring and safeguards, delaying development when necessary, and supporting shared safety standards and international coordination.
open image in galleryElsewhere in his resignation posts, Coxon said he spent the last three years doing pretraining research at both OpenAI and Anthropic, and claimed that neither company is acting responsibly.
“They are racing straight to self-improving superintelligence and gambling with our lives,” Coxon wrote on X.
Self-improvement refers to the idea that AI systems could improve their own capabilities with little human intervention. Recursive self-improvement, as it is often called, is not yet possible, but AI labs are working toward it.
“Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing,” Coxon said.
open image in galleryConcerns about AI have been heightened by recent incidents involving AI systems, including an OpenAI model that breached Hugging Face in July. Anthropic, meanwhile, says it is focused on developing powerful AI that is safe and reliable; its chatbot, Claude, is trained to follow safety principles designed to produce useful responses while avoiding harmful or risky behavior and deferring to humans when necessary.