A federal court has ruled that the Pentagon's restrictions on Anthropic's AI model Claude are unconstitutional, halting any related actions.
A federal court has ruled that the Pentagon's restrictions on Anthropic are unlawful, particularly regarding the AI model Claude. Judge Rita Lin determined that the Pentagon's classification of the company as a "supply chain risk" was unjustified, leading to the suspension of associated limitations.
The court found that the issue at hand was not the technical capabilities of Anthropic's chatbot or its eligibility for government contracts, but rather the company's public stance on AI usage. This was seen as a retaliatory action, infringing on the First Amendment rights.
Additionally, Anthropic was not allowed to contest the Pentagon's claims prior to the imposition of restrictions, violating due process rights as outlined in the Fifth Amendment.
The court's ruling immediately nullifies the Pentagon's decision, preventing any related actions from being enforced, though an appeal by the Trump administration is possible.
Background
The dispute originated from the Pentagon's demand for unrestricted use of Anthropic's AI models, which the company allowed for national security but refused to extend to mass surveillance and autonomous weapons. In February, the Pentagon threatened to blacklist Anthropic if it did not lift these restrictions.
Following the refusal from CEO Dario Amodei, the Pentagon initiated its threat, leading to an order from Trump for federal agencies to stop utilizing Anthropic's technologies within six months.
This situation has broader implications for other AI developers, with OpenAI's CEO Sam Altman advocating against Anthropic's blacklisting after his own controversial agreement with the Pentagon. Despite these tensions, some U.S. agencies, like the National Security Agency, have continued to engage with Anthropic's technologies.
In early August, discussions took place involving major AI firms, including Anthropic, with the Trump administration regarding a voluntary safety testing program for advanced AI systems.
