OpenAI, Anthropic to US: Be Ready to Put Brakes on AI

The companies racing to build the most advanced AI are now publicly asking Washington to be ready to hit the brakes. OpenAI and Anthropic on Tuesday backed a new petition urging the US government to help create international tools to slow "frontier" AI if progress starts moving too fast. Hundreds of employees at those firms—plus staffers at Google, Meta, and others—signed on, reports the Washington Post. Of notable dissent: Mark Zuckerberg, who sat down with the New York Times to discuss his views. A look:

• The ayes: The letter warns that automating AI research could rapidly accelerate progress beyond what can be safely overseen, likening oversight to a software-era version of nuclear-arms control. "It's mostly notable given how many high-up names are on the list. It's increasingly the case that labs and their staff are explicitly talking about international governance," said Andrew Yoon of CivAI, a nonprofit that analyzes AI capabilities and risks. One notable missing name: OpenAI CEO Sam Altman.
• A nay: In a rare interview with the Times, Zuckerberg criticized the closed, safety-first approach favored by leading labs like Anthropic and OpenAI—without naming them—arguing it risks concentrating power and choking innovation. He cast their focus on worst-case scenarios as overly "doom"-driven and out of step with what he describes as American tech's tradition of openness. Zuckerberg instead is championing what he calls "personalized superintelligence" for everyone. "I think it is literally impossible to have a single benevolent superintelligence that is simultaneously aligned with everyone at once," he said.