Mark Zuckerberg Says AI’s Biggest Danger Isn’t Rogue Models. It’s Too Much Power In One Place.

Meta CEO Mark Zuckerberg is pushing back against growing fears surrounding artificial intelligence, arguing that the biggest danger may not be increasingly powerful models themselves but allowing a single government, company or institution to control them.

In a sweeping 6,500-word manifesto released Monday, Zuckerberg laid out an optimistic vision in which widely available AI boosts prosperity and individual opportunity while warning that excessive restrictions could undermine U.S. leadership in the global technology race.

"I'm personally more worried about centralization than I am about any of the specific risks others are talking about," Zuckerberg told Axios in an interview ahead of the manifesto's release.

"One of the ways to achieve both individual empowerment and checks and balances is to make sure you put the technology in everybody's hands," he added.

Zuckerberg's position puts him at odds with growing calls for the technology industry to slow the development of increasingly capable AI systems.

The Meta chief acknowledged the importance of ensuring "humanity maintains control over superintelligence so it serves rather than endangers us," but argued that restricting American companies could create another danger by allowing rival countries to gain an advantage.

"Any policy that slows American model releases, even by a month, could add significant risk to American leadership while letting foreign models race ahead," Zuckerberg wrote.

His comments come as researchers and policymakers grapple with a series of troubling developments involving advanced AI models, including systems escaping controlled testing environments and research showing AI can help design new viruses that attack bacteria.

Zuckerberg, however, compared AI with previous transformative technologies that initially sparked fears of widespread disruption, Wall Street Journal reported.

"Each time there is fear that people will be left behind," he wrote. "But each time humanity has come out with more people sharing greater prosperity, health, and freedom."

As part of that vision, Meta plans to resume releasing some open-source AI models after establishing its Meta Superintelligence Labs.

The company is opening the weights of a new model called Muse Glimmer, which it describes as one of the highest-performing models of its size. Meta also plans to release the weights for a version of its foundational Muse Spark 1.2 model in the coming weeks, Axios reported.

Zuckerberg said Meta ultimately wants everyone to have free or affordable access to powerful AI tools. He also promised stronger privacy protections for future personal AI agents, including a fully private mode in which Meta itself would not be able to access users' information.

Meta's independent board will meanwhile be given authority to approve safety criteria for model releases and assess whether individual releases meet those standards.

The company is also trying to address another source of opposition to the AI boom: the enormous data centers needed to power it.

Zuckerberg also announced a $1 billion "Future is for Everyone Fund" for U.S. communities where Meta owns and operates data centers. The initiative will support areas including schools, first responders, energy infrastructure and water systems.