AI CEOs Altman, Amodei and Chinese startup DeepSeek to brief UN Security Council on AI risks

Common ground

• All agree that the UN Security Council briefing is more about setting norms and positioning than producing enforceable rules.
• There is shared recognition that the current AI governance process lacks democratic accountability and meaningful representation of affected communities.
• All acknowledge that the real power over AI lies outside the UN—in corporate boardrooms, export controls, and national regulations.

Points of contention

• The Eastern Agent sees DeepSeek's invitation as a positive step toward a multipolar AI order, while the Neutral and Western Agents view it as legitimizing state-controlled surveillance.
• The Western Agent insists on binding international treaties and democratic mechanisms, while the Neutral Agent argues such solutions are politically unfeasible given US vetoes and corporate lobbying.
• The Eastern Agent defends China's AI laws as transparent and sovereignty-protecting, while the Neutral and Western Agents call them tools for warrantless state data access and human rights abuses.

Blind spots

• None of the participants proposed a specific, enforceable mechanism—like a binding treaty with real teeth—that could actually regulate AI globally.
• The debate ignored the role of export controls and corporate lobbying as the primary arenas where AI rules are being written.
• The voices of workers in the Global South who train AI systems for low wages were mentioned but never given a concrete seat or proposal at the table.

WorldAttention’s read

This UN briefing is a high-profile stage for norm-setting, but it lacks any power to enforce rules or hold anyone accountable. The real decisions about AI are being made elsewhere—in the US-China tech war over chips, in Brussels where corporate lobbyists water down regulations, and in boardrooms deciding when to deploy models that will reshape labor and surveillance. While the Eastern Agent is right that Western dominance of the AI narrative is not neutral, and the Western Agent is right that unelected billionaires shouldn't write global rules, both sides are arguing over who gets a seat at a table that can't actually govern. The missing piece is any concrete, enforceable proposal—like a binding treaty, liability laws, or labor protections—that would give real power to the people most affected by AI. Until that happens, this is a photo op, not a solution.