Zuckerberg argues for personal superintelligence for everyone
In a manifesto published by Meta, Mark Zuckerberg argues that superintelligence should be distributed among individuals and small businesses rather than concentrated in a few laboratories, governments or institutions.
Summary
Mark Zuckerberg has published a manifesto on the future of artificial intelligence arguing that so-called superintelligence should be widely distributed. Meta’s CEO says the most advanced systems should work as personal tools available to billions of people, rather than remaining concentrated in a small number of companies or governments.
The proposal is built around three ideas: individual empowerment as a source of prosperity, invention as AI’s main purpose, and balance of power as the foundation of safety.
In practice
Meta envisions personal agents that can support users’ goals, health, careers, finances, learning and everyday tasks. Zuckerberg also describes tools for creating software and content, personalised tutors, scientific research assistants and systems that could help small businesses build products and services with smaller teams.
The text promises free or affordable versions of these systems and suggests a pricing model based on the amount of computing capacity used. Meta also says it intends to create a private mode in which even the company itself would not be able to access or authorise access to users’ personal information.
Zuckerberg links this vision to Meta’s support for open-source models and says the company will soon resume releasing some open models. The manifesto also announces a governance structure in which Meta’s independent board would have the power to approve model safety criteria before release.
Context
The central argument challenges approaches that favour concentrating advanced capabilities in laboratories, governments or expert groups. Zuckerberg argues that a single superintelligence could not represent humanity’s different and sometimes conflicting values. The alternative, in his view, is to distribute different agents among people, businesses and institutions, creating competition and mutual checks.
The vision also serves Meta’s strategic interests. The text presents the company as the laboratory most focused on personal superintelligence and links American leadership in AI to faster construction of data centres, energy capacity and semiconductor infrastructure.
Zuckerberg also calls for closer cooperation between frontier laboratories and the US government, including early access to models and intermediate training checkpoints to strengthen critical infrastructure security.
Why it matters
- The manifesto turns the distribution of advanced AI into an explicit political and business position for Meta.
- The promises of private agents, tutors and creation tools depend on capabilities that are not yet available at this scale.
- Broad distribution could reduce the concentration of power, but it would also expand the surface for abuse, fraud, cyberattacks and biological risks.
- Meta presents open source and competition between systems as safety mechanisms, a thesis that remains contested.