Jacob Coxon, Joe Benton quit, raise alarm about fast AI
Technology
Two former Anthropic researchers, Jacob Coxon and Joe Benton, have quit and are now raising serious concerns about how fast AI is advancing.
Benton is now going to join METR Evaluations to independently check AI risks and push for stronger safety rules across the industry.
Benton urges greater AI safety investment
Benton called out AI companies for not investing enough in safety, warning about big risks like intelligence explosions and hidden system failures.
He urged firms to be more open about their progress and report any incidents.
Many at Anthropic share his worries; Benton quoted his ex-boss Evan Hubinger, who believes there's over a 10% chance that AI could cause extinction-level threats if things aren't handled carefully.