Meta CEO Mark Zuckerberg criticized OpenAI and Anthropic in an interview with The New York Times, saying their approach to building AI could centralize power and limit public access to transformative technology. Zuckerberg did not name the 2 rival companies directly, according to the article, but said some companies want to develop AI in a controlled manner and that such an approach would amount to “abandoning our values” in US technology development while stifling innovation.

Zuckerberg said AI development debates have become dominated by warnings about catastrophic risk. He argued for a more open approach and said there should be “a voice or several voices” bringing realism to the debate, while OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have argued that many of the most capable AI models need stronger safeguards because of potential risks.

The dispute comes as OpenAI and Anthropic face scrutiny for lobbying lawmakers in Washington against Chinese open-source models. The article says recent models from Zhipu AI and Qwen have challenged ideas of US supremacy in AI, with leading Chinese models coming close to the benchmark scores achieved by Claude Fable 5 and GPT-5.6 Sol.

Anthropic drew backlash in Silicon Valley after refusing to sign a petition led by Nvidia, Microsoft, Meta and Palantir that asked US lawmakers not to impose “premature restrictions” on open-weight models (models with publicly available parameters). Amodei wrote in a blog post earlier this week that Anthropic is not advocating a ban on open-weight models as a category, and instead supports keeping powerful chips out of authoritarian hands, stopping industrial-scale distillation (copying model behavior at scale), and requiring safety testing for sufficiently capable open and closed models.

Meta CEO Mark Zuckerberg criticized OpenAI and Anthropic in an interview with The New York Times, saying their approach to building AI could centralize power and limit public access to transformative technology. Zuckerberg did not name the 2 rival companies directly, according to the article, but said some companies want to develop AI in a controlled manner and that such an approach would amount to “abandoning our values” in US technology development while stifling innovation.

Zuckerberg said AI development debates have become dominated by warnings about catastrophic risk. He argued for a more open approach and said there should be “a voice or several voices” bringing realism to the debate, while OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have argued that many of the most capable AI models need stronger safeguards because of potential risks.

The dispute comes as OpenAI and Anthropic face scrutiny for lobbying lawmakers in Washington against Chinese open-source models. The article says recent models from Zhipu AI and Qwen have challenged ideas of US supremacy in AI, with leading Chinese models coming close to the benchmark scores achieved by Claude Fable 5 and GPT-5.6 Sol.

Anthropic drew backlash in Silicon Valley after refusing to sign a petition led by Nvidia, Microsoft, Meta and Palantir that asked US lawmakers not to impose “premature restrictions” on open-weight models (models with publicly available parameters). Amodei wrote in a blog post earlier this week that Anthropic is not advocating a ban on open-weight models as a category, and instead supports keeping powerful chips out of authoritarian hands, stopping industrial-scale distillation (copying model behavior at scale), and requiring safety testing for sufficiently capable open and closed models.