Zuckerberg rejects coordinated AI slowdown, says labs can police safety themselves

Meta CEO Mark Zuckerberg has rejected calls for AI labs to jointly slow capability gains, a push led by Anthropic's Dario Amodei and echoed by other lab chiefs. In an X post, he argued each lab already has liability and competitive reasons to build safely, citing Meta's months-long delay of its Muse model for safety work. He favored broader independent testing over coordination.