{"text":[[{"start":10.1,"text":"The Trump administration’s attempt to punish Anthropic for imposing restrictions on the use of its AI in warfare has been blocked by a US court after months of legal wrangling between the Claude maker and the Pentagon."}],[{"start":23.16,"text":"California federal judge Rita Lin ruled on Thursday that the defence department’s attempt to designate Anthropic a supply-chain risk — usually reserved for US adversaries — amounted to “unlawful retaliation”."}],[{"start":36.26,"text":"The decision is a win for Anthropic but still leaves it facing a separate legal tussle in a Washington appeals court, which in April rejected its request to pause the designation. The government could also appeal against the California ruling."}],[{"start":49.54,"text":"The stand-off between Anthropic and the Pentagon began after the start-up earlier this year refused to sign an open-ended contract with the US Department of Defense."}],[{"start":58.84,"text":"Anthropic chief executive Dario Amodei indicated he would stick to two “red lines” prohibiting the use of the company’s technology for lethal autonomous weapons and domestic mass surveillance, enraging defence secretary Pete Hegseth and President Donald Trump."}],[{"start":74.36,"text":"In late February, the administration made the supply-chain risk designation and ordered all federal agencies to stop using Anthropic’s products. The ban, in effect, barred all government defence contractors from doing business with the AI lab, although the action included a transition period during which the Pentagon has continued to use its technology."}],[{"start":93.88,"text":"In her ruling on Thursday, Lin admonished the Trump administration, writing that “the empty invocation of national security is not a blank cheque to punish and retaliate against government critics”."}],[{"start":105.2,"text":"The judgment means that US agencies other than the defence department can continue to use Anthropic’s products. Several agencies have already been testing the company’s powerful Mythos model to analyse cyber and national security risks."}],[{"start":119.32,"text":"Relations between the leading US AI lab and parts of the Trump administration appeared to have improved in recent months, as they worked closely on the release of Mythos, which has been rolled out slowly due to concerns its cyber security capabilities could be used by bad actors."}],[{"start":134.659,"text":"The US government in June briefly blocked another Anthropic model, Fable, on security grounds before reaching a deal with the company to permit its release."}],[{"start":142.72,"text":"Anthropic welcomed the court ruling and said it remained “focused on working productively with the government to harness AI for our national security so all Americans benefit from this technology”."}],[{"start":154.76,"text":"The White House and the Pentagon did not respond to requests for comment."}],[{"start":163.72,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1787888575_9759.mp3"}