He quit, then went viral

Jacob Coxon, 27, left Anthropic and posted about the dangers of AI. The post spread fast, landing in the middle of a growing industry fight over safety.

We could all die

Coxon told the BBC the risk is not abstract or distant. He puts it in the immediate future if the pace of progress holds.

“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”

His old boss agrees

Dario Amodei called for AI development to slow down, saying the risks are serious and governments need time. Sam Altman and Elon Musk both backed the idea.

A race nobody can exit

Coxon says any slowdown has to be coordinated with China or it just becomes an international race. Staff ask for regulation because they cannot stop on their own.

“They find themselves trapped in a race. And they're scared of the outcomes of that race.”

Bots that take the internet

The scenario Amodei raised is a swarm of bots acting as one supercomputer and seizing the internet. Coxon thinks that is realistic in six months to a year.

Buying land, just in case

Coxon says colleagues are planning their lives around this and some are buying land, scared of the instability rapid AI progress could bring.

“They all keep this in their head on a daily basis while working on the technology.”

Anthropic pushes back

The company says it has always flagged both the benefits and the unprecedented risks, publishes its safety testing, and wants a lawful, verifiable way for the industry to pace releases.

More than one in ten

Anthropic scientist Evan Hubinger puts the odds of AI killing everyone above 10% within a decade. Geoffrey Hinton called that figure not unreasonable.

“We really do earnestly believe AI could kill all humans!”

Or it is all hype

Hugging Face's Clement Delangue compared asking Coxon about extinction to asking your AC guy about climate change. Nvidia's Jensen Huang dismissed the comments outright.

Follow the valuations

Critics say doom talk justifies huge valuations and a regulatory moat. Anthropic is reportedly preparing a record IPO, while OpenAI, valued at $852bn, has delayed its own on safety grounds.