A former Anthropic researcher whose warnings about Anthropic researcher AI fears have now reached tens of millions of people online has told the BBC that staff working on artificial intelligence are ‘genuinely frightened’ about where the technology is heading, and that some colleagues are even making contingency plans for civilisational collapse.
Jacob Coxon, 27, left Anthropic after previously working at OpenAI, and published a resignation post that has since been viewed more than 70 million times on X, according to CNBC. He appeared on Sunday with Laura Kuenssberg to explain what drove him to go public with his concerns.
A park bench, a viral post, and a stark warning
Coxon wrote the post that sparked the furore on 8 September, from a park bench in San Francisco’s Alamo Square, according to TIME. The message was blunt: if the pace of AI development is not slowed, the consequences for humanity could be catastrophic.
‘I believe that if we don’t slow down at the current rate of progress, there is a strong chance that we could all die in the immediate future,’ Coxon told the BBC. He said the hardest question to answer was what an AI apocalypse would actually look like, but pointed to one scenario outlined by Anthropic head Dario Amodei: a swarm of bots functioning like a giant supercomputer that could take over the internet. Coxon said that kind of outcome could become realistic within six months to a year.
He also described colleagues at AI firms quietly preparing for a destabilised world. Some, he said, were ‘considering buying land somewhere because they’re so scared of the instability as a result of rapid AI progress’. His peers feared the danger could arrive within the next two years. ‘They all keep this in their head on a daily basis while working on the technology,’ he said.
Industry leaders align on Anthropic researcher AI fears, but Silicon Valley pushes back
Coxon’s departure sits within a broader shift in tone at the top of the AI industry. Amodei published an essay arguing that AI development should slow down, saying the risks were ‘serious’ and that companies and governments must be given time to address them. The bosses of two rival firms (Sam Altman of OpenAI and Elon Musk of xAI) both said they agreed with Amodei’s proposal for industry-wide deceleration, regulation, and independent monitoring of AI model development.
Altman went further, telling Fortune magazine on Friday that an initial public offering for OpenAI would be an ‘ill-advised moment’ given the safety concerns currently in play. Anthropic, meanwhile, is reportedly preparing for a potentially record-setting public offering of its own.
Coxon welcomed the call for a slowdown but said any deceleration would need to be co-ordinated with China to avoid ‘a race at an international scale’. ‘The people who work at these companies are completely serious when they ask for regulation because they find themselves trapped in a race. And they’re scared of the outcomes of that race,’ he said.
The political response has also begun to take shape. Senator Bernie Sanders said he would soon introduce legislation to pause AI development and ban superintelligence, AP News reported, a sign that the debate is moving from tech forums into legislative chambers.
Not everyone is convinced. Nvidia boss Jensen Huang discussed Coxon’s comments at a Goldman Sachs conference and, according to multiple people present who spoke to the BBC, dismissed them as untrue. Huang has previously said the idea that AI is ‘going to be the end of humanity’ is ‘complete nonsense’. The CEO of the AI platform Hugging Face, Clement Delangue, was similarly sceptical, writing on X: ‘Sorry, but asking Jacob about AI extinction risk is like asking your AC guy about climate change.’ He later said he wanted to help develop solutions after Amodei’s essay was published.
Critics have also suggested that warnings about AI peril may be designed to generate hype, or that Anthropic is seeking to trigger a regulatory environment that would leave it and OpenAI with a duopoly.
Anthropic, in response to Coxon’s departure, said it had ‘always been transparent that AI will bring both enormous benefits and unprecedented risks’. A spokesperson told BBC News the firm tests its models ‘aggressively’ and publishes the findings to aid scrutiny, adding that it was the first to publish a framework for mitigating the risks posed by AI development. ‘This work is also why we believe the world would benefit from the industry adopting a lawful, verifiable way to work together to pace how we release powerful models,’ the spokesperson said.
Anthropic scientist Evan Hubinger added his voice to Coxon’s on X, writing that he agreed with the concerns. ‘We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade,’ Hubinger said.
Coxon himself is not without optimism. He told the BBC that the people working on AI ‘genuinely want to see the upside’ of things like ‘solving diseases and improving everyone’s lives’. Whether that optimism survives the race he describes is the question lawmakers on both sides of the Atlantic are now being pushed to answer.
