Zuckerberg Pushes Back on Coordinated AI Slowdown, Says Labs Can Act Alone
Mark Zuckerberg rejected calls for a coordinated slowdown in AI development, saying companies can and should manage safety on their own. He argued that firms have strong incentives to prevent harm because users want trustworthy systems and companies face liability if models cause damage. He cited Meta’s decision to delay shipping Muse for months to focus on safety and security as an example of acting independently rather than waiting for rivals. Zuckerberg said each lab should move at the pace needed to train models safely and take its own precautions. He added that better alignment—making AI follow human intent—would be a competitive advantage because users will avoid systems that don’t obey instructions. He said Meta already uses independent evaluators and wants more outside testing, and that most of Meta’s computing is focused on serving users rather than recursive self-improvement. The comments came after calls from Anthropic, OpenAI, and Elon Musk for slower AI capability gains.
