The Public Should Own Half of the Big A.I. Companies
A proposal by U.S. senator Bernie Sanders to have the public own 50% of major AI companies, via a one-time tax paid in stock into a sovereign wealth fund, has sharply divided opinion. Supporters argue that AI firms built their products on society’s collective data and will drive massive labor displacement, so citizens should share in both the profits and governance of such a transformative technology. Critics counter that seizing equity is unconstitutional expropriation that would chill investment, concentrate even more power in government, and that more conventional tools—taxation, copyright enforcement, and regulation—are better ways to address AI’s risks and externalities.
Rationale for Public Ownership
- Many argue AI models are built on “our collective intelligence” — public, copyrighted, and user‑generated data — often taken without consent or compensation.
- Because AI could be “the most transformational technology” and shapes information, labor, and politics, some see a unique case for public equity, not just regulation or taxes.
- Supporters compare AI firms to oil companies extracting public resources; sovereign wealth funds (Alaska, Norway) are cited as precedents for sharing resource rents.
Property Rights, Taxation, and Legality
- Opponents frame a one‑time 50% stock levy as expropriation/“takings,” beyond normal taxation and likely unconstitutional in the U.S. context.
- Others counter that taxation already seizes significant portions of income; taking equity as tax is seen as a form choice, not fundamentally different.
- There is debate over whether a federal wealth/asset tax must be apportioned and how this interacts with the Takings Clause.
Role and Behavior of AI Companies
- Critics emphasize large‑scale scraping of copyrighted and non‑public data, calling it theft or the “greatest theft in history.”
- Defenders say models are built from publicly accessible information; nothing is removed, only copied, and learning from data (human or machine) shouldn’t be treated as infringement unless courts decide otherwise.
- There is deep disagreement over whether training on copyrighted data is morally or legally distinct from human reading.
Economic and Labor Impacts
- Some expect massive labor displacement, especially in white‑collar and creative fields, and call for mechanisms (sovereign wealth fund, special taxes) to offset social disruption.
- Others argue every major technology shift has both destroyed and created jobs; AI is not unique, and “lump of labor” worries are overstated.
- Skeptics worry public equity stakes would entrench AI bubbles and force bailouts, leaving “the public” holding losses.
Government Control and Governance Risks
- Concerns include politicization of AI, regulatory capture, and using ownership to coerce firms or entrench ruling parties.
- Some propose non‑voting shares to capture profits without political meddling; others insist public ownership must include governance rights.
Alternatives Proposed
- Stronger copyright enforcement and paid licensing instead of equity grabs.
- Broad‑based wealth or corporate taxes applied to all large firms, not AI singled out.
- Public or cooperative infrastructure (municipal broadband–style, digital co‑ops) and labor‑focused reforms (shorter workweeks, stronger safety nets) as fairer responses than nationalizing AI stakes.