DeepMind Researcher Quits Over Google's Military AI Deal
After DHS killings, he fought Google's military deal—and lost.
In a detailed essay titled "Why I Left Google DeepMind," AI researcher TurnTrout recounts his failed internal campaign to prevent Google from signing an unethical military AI contract. He begins with two DHS killings in January 2026, where federal agents shot peaceful citizens, and notes that Google sells Cloud services to DHS agencies. TurnTrout then expanded his efforts to block Google from signing a Pentagon deal that lacked restrictions on killer robots and mass surveillance. He approached senior figures, including Stuart Russell (who promised a statement that never materialized) and Jeff Dean (Google's Chief Scientist, who signed a 2018 pledge against killer robots but did not leverage his influence to stop the deal). Despite a 25-page proposal with contract language and oversight mechanisms, praised by legal experts, the proposal was ignored by CEO Demis Hassabis and senior policy staff. Google ultimately signed a deal with even weaker restrictions than OpenAI's. Unable to stay in good conscience, TurnTrout left. His story highlights the failure of AI ethics commitments under government pressure, as powerful institutions prioritized contracts over pledges.
- TurnTrout's divestment campaign began after DHS officers killed two citizens, linked to Google Cloud sales.
- Jeff Dean and Stuart Russell failed to act despite public pledges against killer robots.
- Google's military deal had weaker restrictions than OpenAI's, and a 25-page counteroffer was ignored.
Why It Matters
This resignation underscores the fragility of AI ethics pledges when facing government contracts and shareholder pressure.