Zuckerberg manifesto pushes an open-source approach on AI as Meta releases its latest model

Meta Platforms on Monday announced the release of a new artificial intelligence model that's open for other software developers to use as CEO Mark Zuckerberg warned of the risks if control of advanced AI is concentrated under a select few companies, institutions or governments.

The tech company included a “permissive” open source license for Muse Glimmer, a lightweight AI model that can run on a personal computer. Zuckerberg said the company would also would provide a way for developers to access a more powerful AI model, Muse Spark 1.2.

Meta has long been a proponent of open-sourcing its AI models, but has struggled to keep up with rivals including OpenAI, Anthropic and Google that have not taken the same approach.

In a lengthy essay published online, Zuckerberg laid out his vision for the future of AI and outlined why he thought open source is a better way for fair development of AI. He also included policy recommendations for keeping the U.S. ahead of rivals, including China.

Zuckerberg warned that if AI “superintelligence” ends up being controlled by a select few individuals, businesses, governments, or even “AI itself, then that will naturally lead to outcomes that are less favorable for everyone else."

The key is finding a balance that favors individuals and making sure that advanced AI is “broadly distributed" to empower billions of people using AI, he wrote.

Zuckerberg also made what could be a veiled reference to Meta's rivals, such as OpenAI and Anthropic, without naming specific companies.

“Most other labs are focused on building AI for companies, governments, or other institutions, so if those labs lead, then the balance of power will favor larger institutions over individuals,” he wrote.

The United States and its allies need to speed up building energy capacity and infrastructure to stay competitive with rivals like China , he said.

Zuckerberg said authorities should also reconsider their position on “ distillation ,” which involves training a less capable model on the outputs of a stronger one. The Trump administration has vowed to crack down on Chinese AI companies using distillation to extract technical features from U.S. AI models. Anthropic earlier this year accused China's DeepSeek of using distillation.

"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," Zuckerberg said. “This is how the world works, and the US will not be able to lead if we restrict ourselves on this front.”