Five Things to Know About Zuckerberg＇s AI Manifesto－By Meghan Bobrowsky, WSJ

Meta Platforms Chief Executive Mark Zuckerberg issued a 6,500-word encyclical laying out the company’s thinking on the artificial-intelligence race , its significance for society and how the U.S. government should regulate and promote the technology. Here are a few key takeaways.

Meta leaning into open source

The company will resume releasing models with open weights, meaning their parameters can be downloaded by anyone, something it paused after reorganizing its AI operations last year. Zuckerberg also argues for allowing distillation, a controversial practice whereby developers query competing models to copy what they know.

“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 writes.

Company will pump more money into towns near data centers

“Sustainable infrastructure development means that communities must benefit significantly from each project,” Zuckerberg writes. He cites the experience of Richland Parish, La., where schoolteachers received $50,000 bonus checks funded by increased economic activity near the company’s Hyperion data-center campus.

The company will launch a $1 billion “Future Is For Everyone Fund” to directly support communities around the facilities—giving itself a new talking point to deploy against the rising tide of antidata-center activism.

Zuckerberg wants government involved earlier in model releases

The Trump administration is in the process of implementing an executive order establishing a voluntary 30-day review period for new models. Zuckerberg has pushed back , saying that is “quite a meaningful amount of time” given the pace of development. His solution: “a close proactive collaboration between frontier labs and the government rather than a rigid process and review timeline that is followed in all cases.”

Artificial-intelligence labs, he says, “should share intermediate training checkpoints of new models for government use and review rather than waiting until training has completed.”

More oversight of model releases within companies, including Meta

Zuckerberg says Meta is adopting a new governance structure that gives its board of directors power to say whether new models adhere to safety criteria the board approves.

“I do not think it is in my, Meta’s, or the world’s best interests for me or anyone else to be a sole decision maker on how superintelligence is deployed,” he writes. He calls for other companies to adopt similar structures, and says there should be “an industrywide version of this process.”

He says people should decide what forms AI takes

Zuckerberg has been using the term “personal superintelligence” to describe Meta’s vision for AI for some time. In his essay, he puts some specifics on it. Users, not the model developer, should be responsible for deciding the system’s values, he says, contrasting that with other companies’ approaches.

He also says there should be “a fully private mode for personal agents where even Meta or any other service provider cannot see or grant access to your information.”