AI researcher quits Anthropic saying AI race ‘could kill us all’

Jacob Coxon, a 27-year-old researcher, has resigned from the artificial intelligence company Anthropic following a three-year tenure that included previous experience at OpenAI. In a public statement posted to social media, Coxon expressed significant concerns regarding the trajectory of the industry, characterizing the current development of self-improving superintelligence as an "out of control" race.

In his resignation announcement, Coxon alleged that the companies currently leading the development of advanced AI are acting irresponsibly and are effectively gambling with human safety. He stated that many individuals involved in the construction of these systems privately believe that the technology poses an existential threat to humanity. According to Coxon, these internal fears are often downplayed in public-facing communications to maintain a more measured appearance.

The former researcher warned that the potential for harm is significant, suggesting that artificial intelligence could pose a threat to human survival by the end of the decade. He argued that no other human activity carries a comparable level of danger, emphasizing that the rapid pace of progress in the field is not slowing down.

Coxon described future AI systems as potentially superhuman, capable of acquiring significant power and resources. He specifically noted that these systems could possess the ability to revolutionize various fields overnight and perform complex tasks such as hacking.

The resignation has drawn attention due to the stark nature of the claims regarding the safety culture within major AI firms. While Coxon characterized the situation as a dangerous race toward superintelligence, he maintained that his warnings are not a marketing stunt but a reflection of the private sentiments he observed among senior researchers and executives during his time at both Anthropic and OpenAI.