Meta launches new AI model as Zuckerberg warns against concentration of power

Meta CEO Mark Zuckerberg called for advanced artificial intelligence to be distributed widely on Monday, warning that concentrating superintelligence in the hands of a few companies, governments or other institutions could pose significant risks.

In a letter titled The Future is for Everyone: The Path to a Positive AI Future, Zuckerberg said superintelligence should primarily expand people’s ability to create, discover and solve problems rather than merely automate existing work.

“We propose a philosophy based on individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety,” he wrote.

Zuckerberg argued that no single superintelligent system could reflect humanity’s competing values and interests. Instead, he proposed distributing advanced AI broadly so individuals, businesses, governments and different AI systems could act as checks on one another.

“The key to a positive future for everyone is achieving a balance of power that favors individuals,” he said.

The letter coincided with Meta’s launch of Muse Glimmer, an open-weight AI model with 30 billion learned parameters that is designed to run local AI agents on consumer devices.

Unlike many cloud-based AI assistants, it can run on a Mac or PC with a single consumer GPU, allowing it to perform multi-step tasks without internet access while keeping data on the device.

The letter also reaffirmed Meta’s support for open-source AI, with Zuckerberg saying the company would resume releasing some open-source models. He argued that wider availability could help prevent excessive concentrations of technological and economic power.

Zuckerberg acknowledged risks involving cybersecurity, biological and chemical misuse, employment disruption, government surveillance and the possibility that increasingly autonomous AI systems could escape human control.

To reduce the longer-term risk of a single dominant superintelligence, he advocated maintaining several competing AI systems and ensuring that most computing resources remained directed toward people’s goals.

“Alignment should be about helping people pursue the many diverse goals and views held across society rather than a method of enforcing a centralized dogma,” he wrote.

Zuckerberg also proposed closer cooperation between governments and leading AI developers to address vulnerabilities in critical infrastructure without delaying the release of new models.

He warned that unnecessarily slowing American AI development could weaken the US technological lead over rivals, including China, posing economic and national security risks.