Tech Firms Lose Veto on Military AI

Google reportedly signed a classified pact that allows the Pentagon to use its AI models for "any lawful" purpose while explicitly disavowing any right to block lawful government operational decisions. The deal includes non‑binding language discouraging domestic mass surveillance and autonomous weapons without human oversight, but those clauses appear not to be enforceable contractual vetoes. — If replicated across providers, such non‑veto agreements shift oversight and accountability for high‑risk AI uses from private companies to the state, raising questions about transparency, enforceability, and democratic control.