{"text":[[{"start":9.4,"text":"The US government has allowed Anthropic to release its newest model to a small group of partners, easing tensions between the Trump administration and the company amid rising industry unease about Washington’s approach to regulating AI."}],[{"start":23.35,"text":"The commerce department gave Anthropic the green light on Friday to put its most advanced model, Mythos 5, back into the hands of 100 or so “trusted partners”, including companies and US government departments, according to people with direct knowledge of the matter. "}],[{"start":38,"text":"The partial release came after Anthropic made progress towards addressing security risks posed by its advanced AI model, the people said. "}],[{"start":46.35,"text":"The commerce department said: “In just two weeks, we have worked diligently to ensure America remains the global leader in AI while safeguarding our security.”"}],[{"start":55.85,"text":"But the model, as well as a rival product OpenAI released on Friday, will not be distributed more widely for now. Instead, the government has asked to vet and approve users, in effect controlling who can and cannot access the technology."}],[{"start":71.2,"text":"It was “bad news” that OpenAI’s model would not be more widely available, the group’s chief executive Sam Altman said in a social media post on Friday. "}],[{"start":79.75,"text":"Altman said providing trusted partners with models first was “reasonable”, but added “this isn’t quite the process that we think is optimal”. "}],[{"start":87,"text":"The White House has set out plans for a voluntary framework to ensure new models are safe before release but is yet to implement an official process, instead intervening on an ad hoc basis to contain AI risks. "}],[{"start":100.45,"text":"Anthropic has been discussing a new policy framework to deal with new model releases during its negotiations with the government, said a person familiar with the matter."}],[{"start":109.25,"text":"The government hit Anthropic with an export control two weeks ago, prompting the AI lab to withdraw its latest models. That export control remains in place, meaning most businesses will not see access reinstated. A publicly accessible version of the model, Fable 5, will not be released at this stage, the people said. "}],[{"start":128.7,"text":"Prominent investors and technologists have warned the government’s approach risks stymieing US development and gives an advantage to adversaries including China. "}],[{"start":138.54999999999998,"text":"Venture capitalist Marc Andreessen in an interview on Thursday said: “We’re in a weird state of the world where the supposedly totalitarian regime is trying to open up the technology, and the supposedly democratic governance system is trying to restrict and control technology.”"}],[{"start":154.79999999999998,"text":"Andreessen has backed multiple AI groups including OpenAI and is an influential figure in Washington. "}],[{"start":160.85,"text":"Dean Ball, a former senior AI adviser to President Donald Trump who will join OpenAI next month, wrote that the administration had created a “de facto involuntary licensing/preapproval regime for frontier models”. "}],[{"start":176.15,"text":"“The administration itself does not seem to know what safety standards or best practices a company would have to observe for them to be comfortable with the broad release of a model that matches or exceeds Mythos in capability,” he added. "}],[{"start":189.8,"text":"Anthropic said it was working to restore access to a “small group of cyber defenders and infrastructure providers” as quickly as possible. "}],[{"start":197.95000000000002,"text":"“We are pleased to see this progress and continue to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again.” "}],[{"start":207.55,"text":"Anthropic’s talks with the government will continue through the weekend, added a person familiar with the matter."}],[{"start":220.3,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1782537047_9389.mp3"}