TechTank Does the Anthropic–Pentagon feud mean the end of responsible AI?
10 snips
Mar 23, 2026 
Valerie Wirtschafter, Brookings fellow on AI and democratic resilience, and Stephanie Pell, Brookings cyber and national security law expert, discuss the Anthropic–Pentagon clash. They cover the refusal to permit lethal autonomous use, legal fights over supply-chain risk, surveillance and de-anonymization concerns, effects on companies doing government work, and what Congress might do next.
AI Snips
Chapters
Transcript
Episode notes
Policy Getting Made In Public CEO Versus Pentagon Fights
- The Anthropic–Pentagon clash exposed that high-stakes AI policy is being decided in adversarial negotiations between CEOs and defense officials.
- Stephanie Pell warned this bypasses Congress and risks making national policy via public fights rather than deliberative lawmaking.
Company Blocks Lethal Autonomy But Not All Military Uses
- Anthropic objected to military use for lethal autonomous weapons but allowed other defense applications like intelligence analysis.
- Valerie Wirtschafter noted Claude was used in planning and reported operational contexts, but Anthropic said it wasn't reliable for targeting decisions without humans.
Mass Surveillance Is Legally Ambiguous But Technically Feasible
- 'Mass domestic surveillance' lacks a precise legal definition, but commercially available data plus AI analytics can produce de-anonymized patterns of life.
- Stephanie Pell explained existing authorities limit collection on U.S. persons but allow purchase and analysis of commercially available data like location records.