Google expands Pentagon's access to its AI after Anthropic's refusal
TLDR
- Google signed a deal giving the DoD access to its AI on classified networks after Anthropic refused similar terms.
Key Facts
- Anthropic declined to allow DoD use of its AI for domestic mass surveillance and autonomous weapons; the Pentagon labeled it a “supply-chain risk.”
- Anthropic and the DoD are in active litigation; a judge last month granted Anthropic an injunction against the designation.
- Google’s contract includes language discouraging domestic mass surveillance and autonomous weapons use, similar to OpenAI’s deal, but enforceability is unclear per the WSJ.
- OpenAI and xAI also signed DoD deals following Anthropic’s refusal; 950 Google employees signed an open letter asking Google not to proceed without similar guardrails.
Why It Matters
- Google’s deal shows other AI labs are moving to fill the gap Anthropic left, even as enforceability of guardrail language remains unresolved.
- The Anthropic lawsuit is the first known case of a U.S. AI company facing a government “supply-chain risk” designation over ethical refusals.
Julie Bort, TechCrunch · 2026-04-28 · Read the original
