Mark Zuckerberg Takes Aim at Anthropic in Debate Over A.I. Slowdown

 

Mark Zuckerberg Takes Aim at Anthropic in Debate Over A.I. Slowdown

 

On social media, Meta’s chief executive appeared to jab at Anthropic by saying that leading A.I. labs should be focused on safety rather than improving their own technology.

 

Eli Tan

By Eli Tan

Reporting from San Francisco

https://www.nytimes.com/2026/09/15/technology/mark-zuckerberg-anthropic-ai-safety.html

Sept. 15, 2026

 

Mark Zuckerberg, Meta’s chief executive, took aim at rival A.I. start-up Anthropic and its chief executive, Dario Amodei, on Tuesday amid a contentious debate over whether to slow down artificial intelligence development.

 

In a social media post, Mr. Zuckerberg said that leading A.I. labs should be focused on safety rather than improving their own technology. He did not name Mr. Amodei or Anthropic, but said that calls to pause A.I. development industrywide were the wrong approach. On Saturday, Mr. Amodei called for a global A.I. slowdown over concerns about the technology’s rapidly accelerating capabilities.

 

“Every lab has the responsibility and incentive to move at the pace required to train its models safely,” Mr. Zuckerberg posted. “Meta has made this commitment and other labs can do this as well.”

 

Debate over the speed and potential dangers of A.I. development has engulfed the tech industry in recent days, fueled by the resignation of a researcher at Anthropic last week who warned that A.I. companies were “gambling with our lives.” Mr. Amodei then spoke up about a slowdown and was joined by Sam Altman, the chief executive of OpenAI; Elon Musk, whose SpaceX rocket company has been ramping up spending on A.I.; and Demis Hassabis, the chair of Google DeepMind.

 

 

Other tech leaders, such as David Sacks, a venture capitalist and adviser to President Trump, have questioned Dr. Amodei’s motivations. They have suggested he was sensationalizing the risks of A.I. to make it harder for other companies to compete with Anthropic and other leading labs.

 

In his post, Mr. Zuckerberg took a middle ground. He acknowledged the risks of A.I., but said companies are already incentivized to keep their users safe and can slow progress on their own. He used last week’s release of Muse, Meta’s new A.I. “agent” that can use software on its own, as an example and said his company delayed the product’s release for several months to make sure it was secure.

 

“We didn’t call for everyone else to do this before we would,” Mr. Zuckerberg said. “We just did it as part of our day-to-day work because it was clearly the right thing for people and for us.”

 

He also jabbed at rival A.I. labs for racing toward “recursive self-improvement,” which is when A.I. is used to develop itself. A.I. companies should instead focus on serving people, he said.

 

Meta has been aiming to catch up in the A.I. race after falling behind Anthropic and OpenAI last year. The company plans to release its most powerful A.I. model to date, internally called Watermelon, in the coming weeks.

 

Eli Tan covers the technology industry for The Times from San Francisco.