Mark Zuckerberg wants to sell the world on personal AI. The harder part is selling the world on who gets to build it.
In a new TechCrunch analysis, Russell Brandom takes apart Zuckerberg’s latest AI manifesto, a 6,500-word argument for the “personal superintelligence” systems Meta is building. The piece notes that the ideas are not entirely new: a version previously appeared in The Wall Street Journal, and Zuckerberg has discussed similar themes on Meta earnings calls. But this appears to be the most detailed public version of the pitch so far.
The pitch, broadly, is optimistic. Zuckerberg is excited about a future where AI becomes personal, abundant, and powerful enough to help people reshape work, creativity, relationships, and daily life. That is the clean version. The messier version is that every time Meta talks about AI as a deeply personal layer, it also reminds people why they are anxious about letting platform companies deeper into their lives.
That is the real story here. The manifesto is not failing because people cannot imagine useful AI. It is failing because the messenger carries the memory of social media’s last grand promise.
Meta Is Asking For Intimacy Before Trust Has Been Rebuilt
TechCrunch’s critique lands because it puts the AI vision next to the social media hangover. Brandom points to a recent survey finding that 64% of Americans believe social media has been harmful to democracy, with a similar share saying it should be more heavily regulated. He also notes that, the same weekend, a court fined Meta $567 million over harms to children. Whether readers see every complaint against Facebook as fair or not, the public mood around Meta is not neutral.
That matters because personal AI is not another feed tweak. It asks for a different level of closeness. A feed predicts what you might watch next. A personal AI assistant wants to know what you are trying to do, what you care about, who you talk to, where you are stuck, and what kind of help might move you forward. It is less about content consumption and more about delegated judgment.
This is where Meta’s framing creates friction. “Personal superintelligence” sounds empowering when described as individual agency. But from a company whose core business was built on behavioral data, recommendation systems, and advertising, the same phrase can sound like a deeper form of dependency. The more useful the AI becomes, the more context it needs. The more context it needs, the more trust becomes the actual interface.
That is also why the AI race cannot be separated from business models. Personal AI will not only answer questions; it will shape discovery, decisions, commerce, and recommendations. That is why the AI story is also becoming an ads and small-business story, as WeRSM recently noted in OpenAI Is Building An SMB Ads Machine. The assistant layer is attractive because it sits close to intent. For users, that can feel helpful. For platforms, it can become the next monetization surface.
The AI Vision Is Bigger Than The Product, And So Is The Suspicion
Zuckerberg’s manifesto appears to be designed as a philosophical reset: a way to position Meta not as a company chasing the next interface, but as a builder of tools that expand human potential. That is a familiar Silicon Valley move. It asks the public to judge the technology by its promised destination rather than by the incentives of the companies deploying it.
The problem is that AI is arriving after two decades of social platforms teaching people to be skeptical of exactly that kind of promise. The public has seen systems introduced as connection tools become engagement machines. It has seen personalization become optimization. It has seen “community” become a business model that often benefits from conflict, compulsion, and scale.
So when Meta talks about abundant superintelligence, many people do not hear abundance first. They hear another system that may become too big, too intimate, and too commercially important to remain user-centered for long.
For brands and marketers, the lesson is not simply that consumers distrust AI. That is too blunt. People will use AI when it is useful, fast, and clearly under their control. The sharper point is that AI adoption will depend on confidence in the surrounding system: who owns the data, how recommendations are shaped, what incentives sit behind the answer, and whether the tool can be understood when it matters.
Meta may be right that personal AI becomes a defining interface. But if the company wants people to bring AI into more private parts of their lives, it cannot treat trust as a communications problem sitting outside the product. Trust is the product condition. Without it, even the most ambitious AI manifesto reads less like a future people can choose, and more like another platform asking to be believed before it has earned the right.