Meta's Zuckerberg reveals AI plans in lengthy manifesto, derides rivals for concentrating power

Meta CEO Mark Zuckerberg published a manifesto on Monday outlining plans to build AI superintelligence for everyone, criticizing rival AI labs for concentrating power, and announced the release of the Muse Glimmer open-weight model for local agent workflows. Zuckerberg also defended AI distillation as a way for the US to lead in AI, as Meta continues heavy investment to catch up with Anthropic, OpenAI, and Google.

Meta META https://finance.yahoo.com/quote/META/ CEO Mark Zuckerberg laid out his vision for AI in a lengthy manifesto https://www.meta.com/thefutureisforeveryone/ on Monday, saying that the company is focusing on building AI superintelligence for everyone, while criticizing rival AI labs, claiming that they are concentrating power in the hands of a few corporations rather than in the hands of individuals. "Rather than centralizing superintelligence, we should distribute it widely and give every person the ability to direct it," Zuckerberg wrote. "This has the potential 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." According to the CEO, Meta plans to provide free access to AI to billions of people, and charge those who want more compute via a "dynamic auction mechanism" that he says 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." As part of that push, Meta debuted its Muse Glimmer open-weight model on Monday. Glimmer, the company says, is designed for use with always-on local agent workflows. In other words, users can download the model and use it to run on their PCs or Macs for agentic use cases. Zuckerberg also criticized AI leaders who warn that AI could have dangerous consequences for humanity. "The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic." Interestingly, Zuckerberg is also pushing for the use of distillation, or using existing AI models to help develop new ones. The practice is controversial as some companies see it as a shortcut for rivals to create their own AI by piggybacking on the time and money others spend creating their own models. But the Meta CEO claims that it is one of the best ways for the US to lead in the AI race. "Some have tried to frame distillation as harmful, but I think it is important to protect the principle that you can learn from anything you can observe," he said. The announcements come as Meta continues to spend billions on its AI buildout, as it seeks to catch up to frontier AI labs like Anthropic ANTH.PVT https://finance.yahoo.com/quote/ANTH.PVT/ , OpenAI OPAI.PVT https://finance.yahoo.com/quote/OPAI.PVT/ , and Google GOOG https://finance.yahoo.com/quote/GOOG/ , GOOGL https://finance.yahoo.com/quote/GOOGL/ , after its Llama 4 models failed to impress developers last year. Since then, the company established its Meta Superintelligence Labs and released a handful of models under chief AI officer and Scale AI founder Alexandr Wang.