{"text":[[{"start":9.95,"text":"JPMorgan Chase has stopped its staff in Hong Kong from accessing Anthropic’s AI models, following a similar move by rival Goldman Sachs in a sign of the intense scrutiny on the technology’s use outside the US."}],[{"start":24.45,"text":"Employees of the Wall Street bank in the Asian financial hub are now unable to access Claude models from a “drop-down” list of approved large language models (LLMs) available internally, according to three people familiar with the situation."}],[{"start":39.2,"text":"One person with knowledge of the move said the decision was based on the wording of Anthropic’s usage terms in its licensing agreement with JPMorgan."}],[{"start":47.5,"text":"Preventing access to the world’s most advanced AI models represents a threat to Hong Kong’s revival as an international financial centre, given their rapid adoption in other parts of the world, particularly for coding."}],[{"start":59.7,"text":"Earlier this year, the FT reported that Goldman had stopped bankers in Hong Kong from using Anthropic. The bank took a strict interpretation of Anthropic’s terms of use that exclude usage in Greater China, including the former British territory, according to one person familiar with the decision."}],[{"start":77.1,"text":"Western AI models such as OpenAI’s ChatGPT and Claude are banned in mainland China as part of the so-called Great Firewall that restricts information from outside the country. "}],[{"start":89.3,"text":"Hong Kong has long operated mostly free of Chinese censorship, with restrictions on usage imposed by the US AI companies themselves. "}],[{"start":98.7,"text":"The models cannot be accessed directly from Hong Kong but international organisations have been able to circumvent geographic restrictions by negotiating global contracts and hosting activity outside China."}],[{"start":110.9,"text":"US AI groups are wary of their models being used in China, partly due to the threat of “distillation”, whereby local actors could train new models through intensive usage of foreign examples. "}],[{"start":123.85000000000001,"text":"Last week, Anthropic was forced to suspend all access to its state of the art AI model Fable after the US government asked it to limit access for foreign citizens, citing national security concerns."}],[{"start":136.5,"text":"Government officials and financial executives had raised concerns over the AI lab’s Mythos model and its potential ability to exploit cyber security vulnerabilities in the global financial system."}],[{"start":148.2,"text":"JPMorgan declined to comment. "}],[{"start":150.54999999999998,"text":"Anthropic did not immediately respond to a request for comment. The company previously told the FT its Claude models had never been officially “supported” in Hong Kong."}],[{"start":166.89999999999998,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1781765232_3651.mp3"}