Zuckerberg's 6,500-word AI manifesto draws criticism for sidestepping real risks

On Monday, Meta CEO Mark Zuckerberg released a roughly 6,500-word essay titled "The Future Is for Everyone," continuing a trend among tech executives of publishing philosophical manifestos. Zuckerberg told reporter Alex Heath that he felt the need to lay out his vision for AI superintelligence before Meta began releasing super-smart models. The core argument is that concentrating AI power in a few institutions is dangerous and that diffusing power through broad access to models is the remedy. Although Zuckerberg does not name OpenAI and Anthropic directly, he takes aim at their approach to closed, controlled models, writing: "Historically hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led to safe or positive outcomes."

WIRED's hosts and contributors were skeptical. They note that Meta's AI strategy has been chaotic, shifting from an initial commitment to open-source and open-weight models to a costly superintelligence lab that has not yet produced competitive results. One host summed up the positioning as a sour-grapes pivot: when a company cannot win a race, it begins to argue that racing is wrong. Zoë Schiffer, a WIRED contributor, points out that the manifesto relies on disarmingly optimistic examples—Zuckerberg describes an AI agent creating personalized recipes so he and his daughter can bake together—while ignoring the documented dangers of powerful AI systems. Schiffer argues the essay "fundamentally miss[es] the fact that these tools are dangerous" and that leaders must take that danger seriously.

The credibility problem is acute. A week before the manifesto's publication, a court ordered Meta to pay $567 million for failing to do enough to protect children's mental health online. One host directly challenges the implied premise: "Why are we trusting Meta with anything, much less AI and its impacts on society?" The hosts also note that Meta announced a new open-weight system called Muse Glimmer on Monday, where users can download and alter the model. The timing and model release underscore what they see as constant pivoting—a company grabbing at the hottest technology without building the institutional rigor to back it up. WIRED ran the essay through Pangram, an AI detection tool, and confirmed it was human-written. One host joked that AI is "good enough for recipes with his daughter, but not good enough to crank out a manifesto."