Mark Zuckerberg has put Meta’s AI ambitions into manifesto form. The Meta CEO published a roughly 6,500 word essay titled "The Future is for Everyone," using it to argue that AI power should not sit inside only a few institutions.
On WIRED’s Uncanny Valley, Brian Barrett, Leah Feiger, and Zoë Schiffer treated the essay less as a settled roadmap than as a revealing signal. Their discussion centered on a simple tension: Zuckerberg is presenting broad access to AI models as a matter of principle, while Meta is still trying to define its own place in the AI race.
A manifesto built around access
The central argument described on the show is that concentrating AI power in a few institutions is dangerous. Zuckerberg’s answer, as the hosts summarized it, is broader access to models as a way to diffuse that power.
That position places Meta in contrast with companies such as OpenAI and Anthropic, which the episode describes as favoring tighter control over private models. Zuckerberg did not name specific companies in the quoted portion of the discussion, but the hosts read the essay as taking aim at that more closed approach.
One line from the essay captured the philosophical frame: "Historically hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led to safe or positive outcomes." The point is clear enough on its surface. If AI becomes too important to society, the argument goes, it should not depend on the choices of a narrow group of private actors.
But the WIRED conversation pushed on whether that principle is enough. A manifesto can describe the world a company wants readers to see. It does not, by itself, prove that the company has the strategy, credibility, or execution to deliver it.
Meta’s AI strategy looks unsettled
Barrett framed the essay as part of Meta’s continuing attempt to catch up in AI. In the episode, he described Meta as having been heavily committed for a long time to open source and open weight models, in part because that approach could counter companies trying to make money more directly from closed models.
That strategy, according to the discussion, had not worked as Meta needed it to. Barrett said Meta then spent billions of dollars on a large hiring push to create a super intelligence lab, but that the lab had not yet produced anything that caught up to where the bigger labs were.
Feiger also pointed to layoffs on the team only a few months earlier. For her, the manifesto could not be separated from that context. The result, in the hosts’ reading, was not a steady march toward one clear AI philosophy, but a pattern of changes in emphasis.
That matters because the essay’s broad message depends on confidence in Meta’s direction. If a company argues that openness is the right answer for society, readers may reasonably ask whether that belief is a long-term commitment or a strategic position shaped by where the company currently stands.
The open-weight push continues
The discussion also noted that Meta announced a new open-weight system called Muse Glimmer on Monday. Feiger said users are able to download and alter the model as they wish.
That detail is important because it connects the essay’s language to a product move. Meta is not only talking about access in abstract terms; it is continuing to put open-weight systems into the world.
Still, the hosts treated the announcement as part of a broader pattern of pivoting. The issue was not simply whether open-weight models are useful. It was whether Meta’s public argument for them is being offered from a position of strength, or from a company still trying to regain ground.
In practical terms, the manifesto raises several linked questions:
- Can broad AI access reduce concentrated power without creating new risks?
- Can Meta make a convincing case for openness while its own AI strategy appears to be changing?
- Will an open-weight system such as Muse Glimmer clarify Meta’s direction, or add to the sense of motion without resolution?
Trust is the harder problem
Barrett also questioned whether Meta is the right messenger for a sweeping reassurance about AI and society. He pointed to a recent court order requiring the company to pay $567 million for failing to do enough to protect kids’ mental health online.
That fact changes the stakes of the manifesto. The essay is not only a technical or competitive argument. It is also an appeal for trust from a company whose social platforms have already faced serious scrutiny over their effects.
The hosts’ skepticism was direct. A long statement about the public good of AI may sound expansive, but readers are likely to judge it against Meta’s record, not only its language. The more ambitious the claim, the more the company has to show that it can manage the social consequences of its products.
This is why the manifesto, as discussed on Uncanny Valley, seemed to leave so much unresolved. Its argument against concentrated AI power is easy to understand. The harder question is whether Meta has earned the authority to present itself as the company best positioned to broaden access responsibly.
What the essay really signals
The most useful way to read Zuckerberg’s AI manifesto may be as a strategic document, not just a philosophical one. It tells the market, competitors, policymakers, and users how Meta wants its AI work to be understood: open, widely available, and opposed to too much control by a small number of institutions.
But the WIRED discussion makes clear that the message lands in a complicated moment. Meta has pushed open source and open weight models, pursued a large super intelligence lab hiring effort, faced layoffs on the team, and introduced Muse Glimmer. At the same time, it is asking people to believe that its approach to AI will serve the public interest.
That is a lot for one essay to carry. A 6,500 word manifesto can frame the argument. It cannot settle whether Meta’s AI strategy is coherent, whether its latest open-weight model will matter, or whether the company can persuade skeptical readers that its vision is more than positioning in a crowded AI race.