Zuckerberg Rejects Global A.I. Slowdown as Meta Readies Watermelon Model in Weeks
Updated
Updated · The New York Times · Sep 16
Zuckerberg Rejects Global A.I. Slowdown as Meta Readies Watermelon Model in Weeks
1 articles · Updated · The New York Times · Sep 16
Summary
Tuesday’s broadside from Mark Zuckerberg argued A.I. labs should manage safety themselves rather than back an industrywide slowdown, countering Anthropic CEO Dario Amodei’s call on Saturday for a global pause.
Meta’s chief said each lab has both the responsibility and incentive to train models safely, citing the company’s own delay of Muse—released last week by several months—to harden security before launch.
The clash follows a week of escalating alarm after an Anthropic researcher resigned and warned A.I. companies were “gambling with our lives,” drawing support for slower development from Sam Altman, Elon Musk and Demis Hassabis.
Meta’s stance lands as it tries to regain ground on Anthropic and OpenAI, with its most powerful model yet—internally called Watermelon—planned for release in the coming weeks.