What Next: TBD | Tech, power, and the future Anthropic vs. the Pentagon
Feb 27, 2026
Sheera Frenkel, a New York Times reporter covering tech and national security, breaks down the Anthropic–Pentagon standoff. Sheera walks through the tense Pentagon meeting, why Anthropic resists uses like mass surveillance and autonomous weapons, and the unprecedented threat to invoke the Defense Production Act. She also covers alternatives, political pressures, and the moral stakes of delegating lethal decisions to machines.
AI Snips
Chapters
Transcript
Episode notes
Pentagon's Two-Edged Ultimatum
- The Pentagon applied contradictory pressure by both threatening to blacklist Anthropic and to compel it under the Defense Production Act.
- Secretary Pete Hegseth demanded a Friday decision and framed Anthropic as either a supply-chain risk or so critical it must be forced to cooperate.
Anthropic's Guardrails Versus Pentagon Access
- Anthropic publicly positions itself as an AI company demanding guardrails, refusing uses like mass surveillance and fully autonomous kinetic weapons.
- That stance clashes with a Pentagon push for unfettered access to AI across military systems.
Anthropic Is Embedded In Pentagon Data Workflows
- Anthropic entered a DoD pilot integrated via Palantir to analyze massive signals intelligence and meeting data.
- Its models proved useful for quickly summarizing and finding patterns across Pentagon datasets.