特朗普政府拟要求美国企业公开披露中文AI模型使用情况
用中文开源模型而不敢声张的美国公司要留意了。Anysphere 的 Composer 2 就是在 Kimi 基础上改的,连 Airbnb 客服都跑 Qwen,国会正推动网站强制披露。如果你的团队在 Qwen 或 Kimi 上搭业务,得提前想好披露口径和合规路径。
特朗普政府正在考虑一项新要求:美国企业需在其网站上公开披露任何中文AI模型的使用情况。这一思路采用"点名曝光"(name and shame)的施压方式,目的是减少对中国模型的依赖,同时避免出台可能激怒美国产业界或北京方面的新限制措施。众议院中国问题特设委员会主席 John Moolenaar 对披露想法表示支持,并主张即使中文模型运行在美国公司自己的基础设施上,风险依然存在。
具体案例显示这种依赖已经相当普遍:Airbnb 的客服智能体运行在 Qwen 上,不过该公司表示其 AI 绝大多数依赖美国模型,中文模型仅通过获得批准的美国本土供应商接入;Anysphere 的 Composer 2 编程模型基于 Moonshot AI 的 Kimi 修改而来,委员会调查人员已向 Anysphere 和 DoorDash 发函,后者将部分内部任务路由至 Kimi。
这种依赖甚至延伸到了政府部门——据 Reuters 9 月报道,一个美国政府网站曾部署 Alibaba 模型来检索拟议中的联邦法规。披露要求与 Moolenaar 的表态共同构成一种信号:托管在美国基础设施上的"防御性说法"正在失去说服力,跨国使用中文模型的透明度压力将持续上升。
The Trump administration is considering making US companies post any use of Chinese AI models on their websites.
The aim is to discourage reliance through “name and shame” pressure, without new restrictions that could anger industry or Beijing.
Airbnb runs its customer service agent on Qwen, yet says its AI runs overwhelmingly on US models and routes Chinese ones only through approved US-based providers.
John Moolenaar (chair of the House select committee on China) welcomed the disclosure idea and argues risks remain even when Chinese models run on US companies’ own infrastructure, which undercuts that hosting defense.
House investigators have also sent letters to Anysphere, whose Composer 2 coding model was built on a version of Moonshot AI’s Kimi, and to DoorDash, which routes some internal tasks through Kimi.
Even a US government website deployed an Alibaba model to search proposed federal regulations, Reuters reported in September.
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison and AI regulation, compliance frameworks, and sovereign compute policy. Key entities and technical topics under continuous tracking include Qwen, Kimi, Moonshot AI, Anysphere.
Source attribution: curated from rohanpaul_ai (@rohanpaul_ai). Cross-source corroboration: 2 independent sources have covered this development (including eric zakariasson), confirming industry-wide signal strength.
- eric zakariasson02:48Source