Zuckerberg urges AI distribution over concentration in lengthy manifesto

On Monday, Meta CEO Mark Zuckerberg released a lengthy manifesto outlining the company's vision for artificial intelligence and its critique of how rivals are approaching the technology. At the core of his argument is a fundamental disagreement about how AI superintelligence should be developed and deployed: Zuckerberg contends that superintelligence should be "distributed widely" and that "every person" should have "the ability to direct it," rather than having it concentrated in the hands of a small number of corporations.

Criticizing what he views as an overemphasis on AI risk by other industry leaders, Zuckerberg wrote that "the notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic." He framed the alternative—distributing superintelligence—as an opportunity to begin "a new era of personal empowerment where individuals can use this powerful new capability to reach their full potential, pursue their interests, and improve their lives and the world more than ever before."

To operationalize this vision, Meta announced several concrete initiatives. The company debuted Muse Glimmer, an open-weight model designed for always-on local agent workflows. Unlike proprietary models accessed through cloud APIs, Glimmer can be downloaded and run directly on users' PCs or Macs, giving them full control over the system. Regarding pricing and access, Meta plans to provide free AI access to billions of people globally, while charging those who require additional compute capacity through a "dynamic auction mechanism." According to Zuckerberg, this mechanism will "guarantee that everyone gets the lowest price possible for the intelligence and compute they're using while also ensuring the capacity is used for whatever people collectively find most valuable."

Zuckerberg also championed the use of distillation—a technique in which existing, often more capable AI models are used to train new ones. While some companies have criticized distillation as allowing rivals to shortcut the research and investment required to build models from the ground up, Zuckerberg defended it as "one of the best ways for the US to lead in the AI race." He argued that "it is important to protect the principle that you can learn from anything you can observe." These announcements come as Meta continues to invest billions in AI development in a bid to catch up to frontier labs including Anthropic, OpenAI, and Google. The context for this push is the disappointing reception of Meta's Llama 4 models among developers last year. In response, Meta established its Meta Superintelligence Labs and has begun releasing models under the direction of chief AI officer Alexandr Wang, who previously founded Scale AI.