A former researcher at Anthropic says some AI workers are becoming increasingly worried about the consequences of rapid advances in artificial intelligence.
Jacob Coxon said researchers are concerned that the technology is developing faster than society can prepare for its potential dangers.
The 27-year-old made the comments after leaving Anthropic and publishing a warning about the future of AI.
Why Coxon left Anthropic
Coxon’s resignation statement focused on the potential dangers of uncontrolled AI development.
He later told the BBC that he believes the industry must reconsider its current pace.
According to Coxon, continuing at the same speed could create a serious threat to humanity.
He said the people working on AI are not simply raising concerns for attention.
Instead, he argued that many are genuinely afraid of where the technology could lead.
Anthropic chief backs a slowdown
Coxon’s comments came as Anthropic CEO Dario Amodei called for a slower approach to AI development.
Amodei has warned that the risks surrounding advanced AI are serious.
He believes governments and technology companies need more time to develop safeguards.
Coxon welcomed the proposal.
However, he said a slowdown would have to involve China to be effective.
He warned that countries could otherwise become locked in an international competition to build increasingly powerful AI systems.
Fear of an AI ‘apocalypse’
Coxon said one of the biggest problems is that nobody can say with certainty what an AI catastrophe would look like.
One scenario involves autonomous AI systems working together on a massive scale.
Amodei has previously raised concerns about networks of AI agents operating with extraordinary computing power.
Coxon said such a development could potentially allow AI systems to take control of parts of the internet.
He believes some of these risks could become realistic within months rather than decades.
Some researchers are preparing for the worst
Coxon said concerns about AI are affecting how some researchers think about their personal futures.
He claimed some workers are considering where they could live if rapid technological change creates widespread instability.
Others, he said, are thinking about the broader consequences of their work while continuing to develop AI systems.
Coxon suggested that these concerns are present among workers on a daily basis.
However, he stressed that AI researchers also want the technology to succeed.
AI could still transform lives
Despite his warnings, Coxon believes artificial intelligence could deliver enormous benefits.
He pointed to the possibility of using AI to help tackle diseases and improve people’s lives.
The challenge, he said, is ensuring that safety develops alongside technological capability.
That has become one of the central questions in the global AI debate.
Experts disagree on extinction risk
Coxon’s concerns have received support from some AI researchers.
Anthropic scientist Evan Hubinger said he believes there is a significant possibility that AI could eventually cause human extinction.
Hubinger has personally placed the risk at more than 10% within the next decade.
Geoffrey Hinton, the computer scientist and Nobel Prize winner often called the “Godfather of AI”, has also warned about the risks posed by increasingly powerful systems.
Hinton recently said that a 10% chance of AI killing all humans was not an unreasonable estimate.
The race continues
The disagreement comes as major technology companies continue competing to develop more capable AI models.
Some industry leaders want stronger safeguards before the technology advances further.
Others believe extreme predictions could distract from the practical benefits AI can already provide.
The debate is therefore moving beyond technology.
It now involves governments, international competition and questions about how much control humans should retain over increasingly powerful AI systems.
For Coxon, slowing the race may be necessary to ensure that the benefits of artificial intelligence do not come at an unacceptable cost.
BBC
