The debate over who controls advanced AI is usually framed in technical terms: alignment, interpretability, capability thresholds. Mark Zuckerberg wants to reframe it as a question of political philosophy. In a Wall Street Journal op-ed published this week, the Meta CEO laid out a vision of personal superintelligence that stands in direct opposition to the concentration-of-power model implicit in how many of his rivals operate.
The central claim is simple: the question is not whether superintelligence arrives, but who gets to use it once it does. Zuckerberg sketches two futures. In one, a handful of institutions hold the most capable systems, deciding unilaterally how they are deployed. In the other, superintelligence is distributed as a tool that individuals control directly, something akin to having a thousand times the cognitive capability available on demand.
Meta commits to building toward the second future, guided by three principles: individual empowerment, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety.
The op-ed takes veiled shots at rival labs without naming them. Zuckerberg writes that it is surprising how much of the discourse from AI developers is filled with doom, and he questions why anyone convinced that AI will eliminate most jobs and much of humanity’s relevance would then rush to build it. He calls the idea that AI danger justifies concentrated power dangerous in itself.
A key analogy runs through the piece: imagine a single person with a superintelligent lawyer in every court case, an unfair advantage even if their position is wrong on the merits. Now imagine everyone has access to the same quality of AI legal counsel. The second scenario, Zuckerberg argues, produces fairer outcomes and a better society.
The economic argument follows a similar logic. He acknowledges that if the balance leans toward automation, the impact on jobs could be negative. But with wide distribution, he expects more jobs, not fewer: startup formation without large capital requirements, and an economy tilted toward small businesses rather than large corporations.
Notably, the op-ed treats different risk categories unevenly. For cybersecurity, Zuckerberg argues that open access improves security over time, citing the history of open-source software. For biological risks, he calls for more coordination between governments and institutions. The essay does not reconcile how a company committed to broad distribution would handle the second category in practice.
The op-ed arrives in the same week as the Pacing the Frontier employee letter, which argues for slowing automated AI research, and as Meta continues to ship its Muse Spark model line broadly. Whether Meta’s product decisions match the philosophy laid out in the op-ed will be the real test of the argument.
Sources: Artificial Intelligence News (Jul 30, 2026); WSJ (Jul 28, 2026); The Verge (Jul 29, 2026)