Executive Summary

The US Department of Defense is ultimatum-issuing the AI company Anthropic to release its technology for military applications. If not fulfilled by Friday, the Pentagon threatens to cancel a $200 million contract, designate the company as a supply chain risk, and activate the Defense Production Act – a Cold War-era law that enables state control over companies. Anthropic has thus far resisted due to security concerns regarding autonomous weapons systems and mass surveillance.

People

Topics

  • Artificial Intelligence & Military
  • National Security vs. AI Safety
  • Defense Production Act
  • Autonomous Weapons Systems

Clarus Lead

The Pentagon is pressuring Anthropic with multiple threats to force access to its AI technology for military purposes. Defense Secretary Pete Hegseth threatens to cancel an ongoing $200 million contract and invoke the Defense Production Act – a Cold War-era law that permits the US government to place companies under state control citing national security. Additionally, Anthropic could be designated a supply chain risk, a blacklist designation normally reserved for supporters of adversaries like Russia or China. The decision deadline: Friday, February 28, 2026.

Detailed Summary

Anthropic is under massive government pressure because it refuses to release its AI systems for military applications. The company justifies its stance with explicit safety concerns: AI models can "hallucinate," generating erroneous content. In a military context – particularly with autonomous weapons systems and mass surveillance – this could result in loss of life. Founder Dario Amodai warned in an essay of the gradual erosion of safeguards in democratic societies.

However, Anthropic stands isolated in this position. OpenAI reformed its policies two years ago to permit military use. Elon Musk's XAI has reportedly already agreed. The Pentagon interprets Anthropic's resistance as a refusal to serve national security interests and therefore deploys existential leverage: The Defense Production Act dates from the Soviet threat era and was last used during the COVID-19 pandemic. Its invocation would effectively force Anthropic to release its technology for "all lawful military purposes" – regardless of internal company policies.

Key Points

  • Pentagon Ultimatum: Release by Friday or face consequences (contract cancellation, control, blacklist status)
  • Fundamental Confrontation: Anthropic defends AI safety against militarization; Pentagon views national security as endangered
  • Industry Trend: Other tech companies are already cooperating with the military; Anthropic is the exception
  • Core Risk: AI-caused errors in autonomous weapons systems could result in civilian casualties

Critical Questions

  1. [Evidence] What specific technical error rates has Anthropic documented that would render autonomous weapons systems unsafe – and has the Pentagon reviewed these analyses?

  2. [Conflicts of Interest] To what extent does the $200 million contract award influence Anthropic's negotiating position, and is there independent oversight of Pentagon pressure tactics?

  3. [Causality/Alternatives] Could Anthropic release its technology under strict oversight conditions (e.g., human-in-the-loop requirements) rather than complete refusal or complete release?

  4. [Mass Surveillance Risk] What specific surveillance scenarios concern Anthropic – and would the Pentagon have accepted transparent use restrictions?

  5. [Precedent] If the Pentagon successfully invokes the Defense Production Act against Anthropic, which other AI companies or tech firms would be at risk?

  6. [Legitimacy of Threats] Is the "supply chain risk" designation (typically for adversary-country connections) legally appropriate for corporate security concerns?

  7. [Enforceability] How would the Pentagon technically monitor compliance with military use and ensure that Anthropic does not simply circumvent usage restrictions?

Further Reports

  • OpenAI & XAI Cooperation: Both companies have already granted military access; the industry trend points toward stronger defense-tech integration in the AI sector.

Source Directory

Primary Source: SBS German Podcast – 26.02.2026 – sbs.com.au

Verification Status: ✓ 26.02.2026

This text was created with the support of an AI model.
Editorial Responsibility: clarus.news | Fact-Check: 26.02.2026