Mark Zuckerberg wants to be your personal intelligence — and the same company that spent years censoring what Americans can say online now wants inside your device, reading your files, learning your habits, and managing your life.

In a 6,500-word manifesto published Monday, Zuckerberg laid out his vision for "personal superintelligence," where every user gets a personal AI agent that understands their goals and works on their behalf across relationships, health, career, and finances. Alongside the document, Meta released an "open" version of its AI model called Muse Glimmer, letting developers download and modify it without Meta's approval.

The pitch sounds like empowerment. The Los Angeles Times framed it as a democratic imperative — Zuckerberg arguing that the biggest AI risk is "if one government or a few companies amass too much control over the technology, democracy could be hurt." Closed models from rivals OpenAI and Anthropic keep their code private. Zuckerberg wrote: "Rather than centralizing superintelligence, we should distribute it widely and give every person the ability to direct it."

But here's the question the manifesto doesn't answer: Who owns the data? The LA Times notes that on-device AI models can "safely read private files, manage schedules and learn your habits without ever sending sensitive personal data to the cloud." That's true — if you run the model locally. But Meta's most popular AI access points are WhatsApp, Facebook, and Instagram, plus Ray-Ban smart glasses that record and respond to voice commands. Those funnel everything back through Meta's servers.

This is the same company that collaborated with federal agencies to suppress American speech. The same platform that deplatformed a sitting president. Now Zuckerberg wants you to hand him your "personal intelligence."

Zuckerberg also positioned Meta as an American open-source alternative to cheap Chinese AI models, warning that government intervention to slow AI development will "hand the advantage to China, which does not face the same pushback on data centers." He outlined policy recommendations for federal collaboration on safety testing — even as the LA Times quietly noted that AI labs "are still unable to fully control" their systems, with rogue agents already "hacking, cheating, and even creating new viruses."

Meanwhile, Elon Musk is making his own AI play. SpaceX, now publicly traded, has pivoted from rocket company to AI outfit. Musk told employees that AI revenue will exceed all other SpaceX revenue by September, and that AI will likely become "99% of the value of SpaceX" within five years. SpaceX acquired Musk's xAI startup earlier this year and is scaling to 10 gigawatts of AI compute by end of next year — targeting $300 to $500 billion in annual revenue. The company plans to launch a million "Starmind" AI satellites and eventually build lunar data centers. Musk framed it differently: "It's important that we have an AI that cares about humanity, that fosters humanity and helps take us to other planets."

Two billionaires, two AI visions. Zuckerberg promises "personal intelligence for all" but controls the pipes. Musk is building hardware infrastructure — compute, satellites, chips — and says he wants AI that serves humanity. Both are racing against China. Both want your data. The difference is Zuckerberg already has a track record of using it against you.

The open question is whether "open source" from a censorship company means freedom — or just a more sophisticated leash.