AI staff 'genuinely frightened' for humanity's future, ex-Anthropic researcher tells BBC

NEW: A former Anthropic researcher tells the BBC that AI staff are 'genuinely frightened' for humanity's future, warning of a 'strong chance we could all die' if progress isn't slowed, as the firm's boss urges deceleration over 'serious' risks.

An AI researcher quit Anthropic saying there's a strong chance we could all die.

His name is Jacob Coxon, he's twenty seven, he worked at OpenAI before Anthropic, and he told the BBC the people building this technology are genuinely frightened. You can hear him here. He says staff at these companies are completely serious when they beg for regulation, because they're trapped in a race and scared of where it ends.

His old boss Dario Amodei wrote this weekend that the risks are serious and development needs to slow down. And what caught my eye, Sam Altman and Elon Musk both agreed with him. When the three men racing each other all ask for a speed limit, you pay attention.

Coxon says a slowdown only works if China is in it too, otherwise the race just goes international. And his timeline for one scenario, a swarm of bots taking over the internet like a supercomputer, was six months to a year. Six months.

This is Liu Yang, reporting for HIT.