Anthropic just won its courtroom showdown with the Pentagon. U.S. federal Judge Rita Lin ruled that the government’s decision to brand the company a “supply chain risk,” order federal agencies to stop using Claude, and restrict defense contractors from working with it was unlawful. The court found the measures amounted to retaliation for Anthropic’s public criticism of government AI policy.
The ruling is a serious takedown of the government’s evidence. The entire basis for the measures was a four-page memo — and that document came after two of the three actions were already taken. The government had also worried Anthropic could modify its models remotely through a “backdoor” after deployment, but court filings confirmed Anthropic has no such access. In the judge’s view, what’s left is plain distrust of Anthropic and a desire to make a public example of the company.
The clash traces back to Claude’s military-use restrictions. Anthropic has refused to let its model be used for mass surveillance of Americans or fully autonomous weapons; the Pentagon wanted access for all lawful purposes. The court also made clear the ruling doesn’t force the military to keep using Claude. The Pentagon can still switch vendors lawfully — it just can’t rely on the “supply chain risk” label to freeze Anthropic out.
Source: Bloomberg
For more: Dongcha Beating channel · Dongcha Beating discussion group