独家:特朗普政府权衡强制美国企业披露中国AI使用情况记者 Alexandra Alper 华盛顿11月18日电 - 据三名知情人士透露,特朗普政府正在考虑要求美国公司公开披露其对中国AI模型的使用情况,旨在不实施可能激怒业界或北京的新限制措施的前提下,劝阻对该技术的依赖。
Exclusive-Trump administration weighs forcing US firms to disclose Chinese AI use By Alexandra Alper WASHINGTON, Nov 18 - The Trump administration is considering requiring US companies to publicly disclose their use of Chinese AI models, a bid to discourage reliance on the technology without imposing new restrictions that could anger industry or Beijing, three people told Reuters.
知情人士称,财政部和白宫科技政策办公室(OSTP)已讨论该计划,该计划将适用于Airbnb等公司。该公司使用中国阿里巴巴开发的AI模型“通义千问”来驱动其AI客服代理。
The Treasury Department and the White House Office of Science and Technology Policy (OSTP) have discussed the plan, which would apply to companies such as Airbnb, the people said. The company uses Qwen, an AI model made by China's Alibaba, to power its AI customer service agent.
根据该计划,Airbnb等公司将被要求在其网站上发布该信息。
Under the plan, companies like Airbnb would be required to post that information on their websites.
目前尚不清楚这一想法在政府内部是否正获得支持,或者如果确实推进,是否会通过正式的规则制定程序予以确立。
It was not immediately clear if the idea was gaining traction within the administration or if it would be enshrined in a formal rulemaking process if it does move forward.
但其被纳入考量(首次由路透社报道)凸显了政府内部正在兴起一股力量,旨在遏制功能强大的中国开放权重模型的扩散,这在政府内部引发了深切担忧。开放权重模型将其核心组件公开提供,供任何人下载和改编。
But its consideration, first reported by Reuters, underscores an emerging push within the administration toward combating the spread of powerful Chinese open weight models, a source of deep concern within the government. Open weight models make their core components publicly available for anyone to download and adapt.
该提议还显示,官员们正在寻找通过“点名羞辱”运动劝阻使用中国AI的途径,同时坚持在AI问题上更广泛的反监管立场。
The proposal also shows officials are looking at ways to discourage use of Chinese AI through a "name and shame" campaign while sticking to a broader anti-regulation stance on AI.
白宫、财政部、阿里巴巴和中国驻华盛顿大使馆均未立即回应置评请求。Airbnb拒绝置评。
The White House, the Treasury Department, Alibaba, and the Chinese Embassy in Washington did not immediately respond to requests for comment. Airbnb declined to comment.
这一考量之际,关于英伟达CEO黄仁勋与一名涉嫌中国间谍的婚外情的披露,在华盛顿引发了对华采取更强硬立场的新呼声。这使聚光灯聚焦于英伟达AI芯片被走私至中国却未受惩处的案例,加大了对特朗普政府打击中国技术威胁的压力。
The consideration comes as revelations about Nvidia CEO Jensen Huang’s affair with an alleged Chinese spy have fueled renewed calls in Washington for a tougher stance on China. They have turned a spotlight on cases of Nvidia AI chips being smuggled to China that went unpunished, dialing up pressure on the Trump administration to crack down on Chinese tech threats.
中国制造的开放权重模型正在美国企业中获得青睐,因为它们比美国顶尖实验室OpenAI和Anthropic的专有AI工具更便宜,且性能足够好。
Chinese-made open-weight models are gaining ground with US companies because they are cheaper than proprietary AI tools from top US labs OpenAI and Anthropic, and perform well enough.
但对华强硬派担心,北京正利用此类模型——如Moonshot AI(月之暗面)的Kimi、DeepSeek或Qwen——吸引美国企业,使其对中国AI产生依赖。
But China hawks fear Beijing is using such models — like Moonshot AI's Kimi, DeepSeek or Qwen — to hook US companies so they will become dependent on Chinese AI.
美国财政部长斯科特·贝森特7月表示,如果发现中国AI公司通过蒸馏美国模型窃取知识产权,特朗普政府可能会对其实施制裁。蒸馏是指利用大模型的输出训练较小AI模型的过程,旨在降低训练强大新AI工具的成本。
US Treasury Secretary Scott Bessent said in July that the Trump administration could sanction Chinese AI companies if they were found to be guilty of intellectual property theft via distillation of American models. Distillation is the process of training smaller AI models using the output of larger ones as part of an effort to lower the costs of training a powerful new AI tool.
贝森特的言论呼应了白宫科技政策办公室(OSTP)主任迈克尔·克拉齐奥斯4月的评论,后者指责中国在“工业规模”上窃取美国人工智能实验室的知识产权,这促使国务院牵头向外国同行强调此类行为。国会调查 但美国官员迄今尚未限制或惩罚中国AI公司。
Bessent's remarks echoed comments in April by Michael Kratsios, OSTP director, who accused China of stealing US artificial intelligence labs' intellectual property on an "industrial scale," spurring a State Department-led push to highlight such behavior to foreign counterparts. CONGRESSIONAL INVESTIGATION But US officials have stopped short of restricting or taking punitive action against Chinese AI companies.
该问题已引起国会关注。共和党籍众议员、纽约州的安德鲁·加巴里诺和密歇根州的约翰·穆莱纳尔分别担任众议院国土安全委员会主席和众议院对华特别委员会主席,他们于今年启动了对美国企业使用中国AI情况的调查。
The issue has attracted the attention of Congress. Republican US Representatives Andrew Garbarino of New York and John Moolenaar of Michigan, who chair the House Homeland Security Committee and the House select committee on China, respectively, launched an investigation into American companies' use of Chinese AI this year.
他们向Airbnb以及Anysphere发送了信函,后者是AI编程工具Cursor的制造商,其Composer 2 AI模型基于Kimi的一个版本构建。一封信函也发给了DoorDash,该公司联合创始人曾表示,公司在内部系统中通过Kimi处理部分任务。
They sent letters to Airbnb as well as Anysphere, the maker of the AI coding tool Cursor, whose Composer 2 AI model was built on a version of Kimi. A letter was also sent to DoorDash, whose co-founder said the company routes some tasks in an internal system through Kimi.
Moonshot(月之暗面)、DoorDash和收购了Anysphere的SpaceX均未立即回应置评请求。
Moonshot, DoorDash, and SpaceX, which bought Anysphere, did not immediately respond to requests for comment.
中国AI模型在10月成为OpenRouter上按Token使用量计算的三大最热门模型。OpenRouter是一个追踪使用数据的AI市场平台。Token是AI模型处理的数据单位。
Chinese AI models were the three most popular by token usage in October on OpenRouter, an AI marketplace platform that tracks usage data. Tokens are units of data processed by AI models.
据路透社9月报道,中国AI模型甚至被美国政府自身采用。一家美国政府网站部署了中国阿里巴巴的AI模型,以便用户搜拟议中的联邦法规,尽管联邦调查局指控该模型的开发者复制了Anthropic的技术。
Chinese AI models have even been used by the US government itself. Reuters reported in September that a US government website deployed an AI model from China's Alibaba to allow users to search proposed federal regulations even though the FBI accused the model's developer of copying Anthropic's technology.
美国开放权重模型包括Meta的Muse Glimmer、英伟达的Nemotron以及Reflection AI的Beam。
American open-weight models include Meta's Muse Glimmer, Nvidia's Nemotron and Reflection AI's Beam.