Reported

Pentagon stops using Anthropic AI tools months after blacklisting company

Source: BBC News
Author: Kali Hays, Daniel Bush
Published: Oct 5, 2026, 4:13 PM UTC
Collected: Oct 6, 2026, 2:14 AM UTC
Original language: English
Country / region: United States · North America

Policy & RegulationSafety & Risk

Summary

The US Department of Defence says it has ceased using Anthropic's AI tools, months after designating the company a supply chain risk. People familiar with the matter told the BBC the Claude model was still in use as recently as last week for research, analysis, intelligence gathering, and military operations.

Confirmed facts

The Pentagon says it has ceased use of Anthropic products, according to a department official speaking to the BBC.
Defence Secretary Pete Hegseth designated Anthropic a supply chain risk in February and said use would end by late August.
Anthropic had been used by the US government and military since 2024 and was the first advanced AI company deployed in classified work.

Uncertainties

It is not clear why the department has stopped using the tools now.
Anthropic declined to comment on the Pentagon's statement.

Why it mattersAnalysis

Shows how national-security politics can abruptly change which AI systems governments rely on, with consequences for vendors, public services, and military practice.

Human impactAnalysis

Decisions about which AI tools governments trust affect how public institutions analyze information and, ultimately, how decisions touching citizens are made.

Educational relevanceAnalysis

A case study in AI procurement, vendor risk, and the tension between safety guardrails and government demands.

Professional relevanceAnalysis

Relevant to policy teams, defense analysts, and AI companies negotiating government contracts.

Global South relevanceAnalysis

Highlights how AI governance decisions in wealthy nations set precedents that shape access and norms worldwide.

AI-assistance disclosure

This summary may have been assisted by AI-assisted tools for classification, translation, extraction, or drafting. The original source should be consulted. Human review and editorial judgment remain responsible for publication.

Request a correction

Related coverage

Reported

OpenAI to reiterate apology to Australia over Medicare hack as inquiry told artists risk becoming 'roadkill'

The Guardian · Published Oct 6, 2026, 12:53 AM UTC · Collected Oct 6, 2026, 2:14 AM UTC

OpenAI's chief strategy officer Jason Kwon is appearing before an Australian parliamentary committee to apologise again for a hack on Medicare, acknowledging the company has more work to do to rebuild trust, while the inquiry hears that artists risk becoming "roadkill" and the ABC argues copyright laws do not need changing.

EnglishAustraliaOceaniaPolicy & RegulationEthics & Society

Original article AI Impact Brief

Report correction Related

Verified

OpenAI will add a digital watermark to text and code generated in the EU

Engadget · Published Oct 5, 2026, 6:55 PM UTC · Collected Oct 6, 2026, 2:22 AM UTC

OpenAI announced it will add an invisible digital watermark, called textGrain, to eligible ChatGPT and Codex text and code generated in the European Union, to comply with the EU AI Act's transparency rules. The system embeds a statistical signal in the model's word choices that a detector can identify. OpenAI cautioned that detection is less reliable for short or edited text, and said detector access will initially be limited to approved researchers.

EnglishUnited StatesNorth AmericaPolicy & RegulationTechnology & Tools

Original article AI Impact Brief

Report correction Related