Mark Zuckerberg is pushing back on one of Silicon Valley’s loudest calls to slam the brakes on artificial intelligence. In a social media post on Tuesday, the Meta CEO criticized the idea of a blanket slowdown on AI development—a proposal championed days earlier by Anthropic chief Dario Amodei—arguing that labs should instead move at a speed that lets them train and release models safely on their own, reports the New York Times. “Every lab has the responsibility and incentive to move at the pace required to train its models safely,” Zuckerberg wrote, saying Meta had delayed its new AI “agent,” Muse, for months to run extra security checks.
Zuckerberg’s comments drop into an increasingly tense industry fight over how fast AI should advance, sparked in part by an Anthropic researcher’s recent resignation and warning that companies were “gambling with our lives.” Amodei’s slowdown call has drawn support from OpenAI’s Sam Altman, xAI’s Elon Musk, and Google DeepMind’s Demis Hassabis—as well as skepticism from some tech investors who suggest the move could entrench leading labs.
Zuckerberg has staked out a middle position: acknowledging AI’s risks while rejecting a coordinated pause, and taking a swipe at rivals for chasing “recursive self-improvement” instead of building tools that directly serve users. Meta, still trying to catch Anthropic and OpenAI, plans to roll out its most advanced model yet, code-named Watermelon, in the weeks to come.
Reuters notes that President Trump has also expressed skepticism at pausing AI development, noting there are guardrails in place and that any doubt expressed over AI would give China a leg up. Another big name who agrees that China would gain an advantage with a mass AI slowdown in the US: Nvidia CEO Jensen Huang.