# Defense Department Presses Anthropic on AI Safeguards

The U.S. Department of War’s negotiations with Anthropic over a potential **$200 million** AI contract collapsed in early 2026 after Anthropic declined to remove safeguards against using its models for fully autonomous lethal weapons and mass surveillance. Separate defense AI awards announced in summer 2025 provided up to $200 million each to Anthropic, Google, OpenAI, and xAI; the signals do not confirm a new $2 billion award or solicitation. A federal-use directive or supply-chain-risk designation affecting Anthropic could also reach contractors’ commercial delivery chains, including federal systems that use Claude through Palantir’s Maven Smart System.

- Contractors should identify where Claude or other Anthropic services are embedded in federal and commercial offerings, including through subcontractors and systems such as Maven, and assess potential continuity impacts.
- The reported prior awards establish a multi-vendor defense AI market, but the failed negotiations signal that permitted uses and vendor safeguards may affect future sourcing decisions.
- Because the signals report differing court rulings and potential effects from a federal-use directive or risk designation, organizations with Anthropic dependencies should account for that uncertainty in their delivery planning.

**Jurisdictions:** federal
**Industries:** Defense & Military, Information Technology
**Topics:** Artificial Intelligence, Regulatory Compliance
**Published:** October 03, 2026

### Government Entities
- U.S. Department of War (DoW)
- U.S. Department of Defense (DoD)
- U.S. federal agencies
- U.S. Court of Appeals for the D.C. Circuit (D.C. Circuit)

### Vendors
- Anthropic (Awardee of a prior defense AI contract; prospective contractor in negotiations that collapsed)
- Palantir Technologies (Provider of the Maven Smart System, through which contractors may use Claude)
- Google (Awardee of a prior defense AI contract)
- OpenAI (Awardee of a prior defense AI contract)
- xAI (Awardee of a prior defense AI contract)

### Key Quotes
> However, in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values.
> — Dario Amodei, Chief Executive Officer of Anthropic

### Sources
- [Palantir Bulls Pile In As Anthropic Feuds With Trump's Department Of War Over AI Safety — But There’s A Catch](https://stocktwits.com/news-articles/markets/equity/palantir-bulls-pile-in-as-anthropic-feuds-with-trump-s-department-of-war/cZTcJfKRIRt) - Stocktwits
- [Anthropic warns Pentagon fight could cost it billions beyond defense](https://cryptobriefing.com/anthropic-pentagon-dispute-government-pressure) - Crypto Briefing