Win for Anthropic: The Pentagon had no legitimate reason to declare them a security risk

Judge Rita Lin’s decision, dpa reported, represents an important victory for Anthropic in the dispute with the US Department of Defense over the use of artificial intelligence in the military. The Pentagon labeled the company a national security risk after it refused to allow the use of its AI models for mass surveillance of American citizens and autonomous weapon systems.

Anthropic instead sought restrictions on such use, while the Pentagon demanded the ability to use the technology for all lawful purposes.

Judge Lin concluded that the Pentagon’s decision was illegal and unfounded, and that its goal was to punish Anthropic for criticizing the government, not to address a real security risk. Earlier this year moreć had temporarily blocked the implementation of the Pentagon’s decision.

The security risk designation could have had consequences beyond cooperation with the US military. President Donald Trump ordered federal agencies to stop using Anthropica’s technology, while Defense Secretary Pete Hegseth said companies that want to do business with the US military would have to stop working with Anthropica.

The court, however, emphasized that the Pentagon is not obligated to use Anthropic’s technology and that it can switch to the products of other companies for artificial intelligence, provided that it does so in accordance with the law. The Pentagon is moreć has reduced its use of Anthropic’s technology and plans to phase it out entirely by the end of September.

The government has not yet confirmed whether it will appeal the decision. Marking Anthropic as a security risk was an unusual move, because the word about an American company, and so far such designations have mostly referred to foreign companies.

Anthropic CEO Dario Amodei warned that artificial intelligence can automatically collect and link data about individuals available on the Internet and thus create a detailed picture of their lives. He also pointed out that AI is not yet reliable enough for fully autonomous armed systems and said that the company will not knowingly offer technology that could endanger American soldiers or civilians.