内容来源：https://www.wired.com/story/ftc-removes-blog-posts-about-ai-authored-by-by-lina-khan/

内容总结：

2024年7月下旬，美国联邦贸易委员会（FTC）时任主席莉娜·汗在旧金山创业加速器Y Combinator的活动上发表演讲，公开表态支持开源人工智能发展。此次演讲恰逢加州议会审议具有里程碑意义的SB 1047法案之际，该法案拟对人工智能企业实施新的测试与安全要求，后遭州长纽森否决。

汗在演讲中呼吁采取更宽松的监管路径，强调开源模型可使"小型企业将创意推向市场"。其团队同期在FTC官网发布题为《论开源权重基础模型》的博客，提出以"开源权重"替代"开源"的精准表述，特指公开训练权重的AI模型。值得注意的是，特朗普政府今年9月已悄然撤下该博客，并移除汗任期内的多篇人工智能政策文章，包括2025年1月发布的《人工智能与消费者侵害风险》等重要文献。

此次内容清理引发法律合规质疑。前FTC公共事务主任道格拉斯·法拉尔指出，作为AI市场关键监管机构，删除政策博客的行为令人震惊。尽管特朗普政府7月发布的《人工智能行动计划》明确支持开源模式，但FTC大规模删除前任政策的举动仍被业界解读为监管转向信号。

据调查，今年以来FTC已删除汗任内发布的数百篇行业指南与政策博客，涉及人工智能监管、科技巨头反垄断诉讼等核心议题，其中多篇曾获专业机构奖项。目前汗本人撰写的200余篇政策文件仍保留于官网，包括2024年对涉嫌欺诈性AI计划的执法记录等重要文献。

中文翻译：

2024年7月下旬，时任美国联邦贸易委员会主席的莉娜·汗在旧金山创业加速器Y Combinator举办的活动上发表演讲，将自己定位为开源人工智能的倡导者。此次活动举行之际，加州立法者正在审议一项名为SB 1047的标志性法案，该法案拟对人工智能公司实施新的测试与安全要求。这项后来被加州州长加文·纽瑟姆否决的法案遭到批评人士反对，他们认为该法案会阻碍开源人工智能模型的开发与发布。汗呼吁采取限制性较低的方式，并指出由于开放模型可供使用，"规模较小的参与者能够将其创意推向市场"。

在该活动举办前的数日，汗的团队成员在联邦贸易委员会官网发布博客文章，阐述了类似观点。文章指出"开源"一词已被用于描述具有各种不同特征的人工智能模型。作者建议改用"开放权重"这一术语，指代公开训练权重的模型，允许任何人检查、修改或重复使用。

据两位知情人士向《连线》杂志透露，特朗普政府此后已撤下该博文。互联网档案馆的Wayback Machine显示，这篇2024年7月10日发布、题为《论开放权重基础模型》的联邦贸易委员会博文，已于今年9月1日被重定向至该机构技术办公室的登录页面。另一篇由联邦贸易委员会两位技术专家于2023年10月撰写的《消费者对人工智能表示担忧》同样被重定向至技术办公室页面。据Wayback Machine记录，此次重定向发生在今年8月下旬。

由汗的团队撰写并于2025年1月3日发布的第三篇关于人工智能的博文《人工智能与消费者权益受损风险》，如今跳转至显示"页面未找到"的错误屏幕。Wayback Machine记录表明，截至8月12日该博文仍存在于联邦贸易委员会网站，但在8月15日前已被移除。原文中，汗的团队曾写道该机构"日益关注人工智能在现实世界中造成损害的潜力——从助长商业监控到促成欺诈和假冒行为，再到固化非法歧视"。

这些博文被移除的原因尚不明确。联邦贸易委员会发言人未回应置评请求。汗通过发言人表示不予置评。

前联邦贸易委员会公共事务主任道格拉斯·法拉向《连线》表示，鉴于该机构作为人工智能市场关键监管者的角色，开放权重博文被撤尤其令人惊讶。"看到弗格森领导的联邦贸易委员会在此次向市场释放的信号中与特朗普白宫如此不一致，我感到震惊。"他提及新任命的联邦贸易委员会主席安德鲁·弗格森时如是说。

特朗普政府于7月发布的《人工智能行动计划》主张："我们需要确保美国拥有基于美国价值观的领先开放模型"，且"联邦政府应为开放模型创造支持性环境"（联邦贸易委员会未回应这些删除行为是否代表政策转变的质询）。

包括白宫人工智能与加密技术特别顾问戴维·萨克斯、白宫人工智能高级政策顾问斯里拉姆·克里希南在内的多位特朗普政府科技顾问，也始终倡导开源人工智能，将其视为美国维持技术优势的关键途径。

据《连线》此前报道，自特朗普总统于今年1月重返白宫以来，联邦贸易委员会已移除汗任内发布的数百篇针对科技行业的博文及商业指南。3月份，该机构删除了约300篇涉及人工智能、消费者保护及其对亚马逊、微软等科技巨头诉讼的帖子。其中获奖博文《诱惑测试：人工智能与消费者信任的构建》曾为科技公司提供避免开发欺骗性AI聊天机器人的指导。

联邦贸易委员会内部人士今年3月向《连线》透露，删除公开博文"根据《联邦记录法案》和《开放政府数据法案》可能引发严重合规问题"，这些法律要求政府机构保存具有行政、法律或历史价值的记录并向公众开放。拜登政府时期，联邦贸易委员会领导层曾在持异议的往届政府商业指令和指南上添加"警告"标签。

截至发稿时，由汗亲自撰写的200余篇文稿仍保留在联邦贸易委员会官网，包括2024年9月关于打击涉嫌欺诈性AI计划的执法行动博文、2024年联邦贸易委员会关于生成式AI基础模型市场竞争的联合声明，以及她在2023年生成式AI圆桌会议上的发言。汗在当时指出该机构正"密切关注人工智能如何加剧欺诈行为"并"巩固控制关键原材料的企业主导地位"。

英文来源：

In late July 2024, Lina Khan, then the chair of the US Federal Trade Commission, gave a speech at an event hosted by the San Francisco startup accelerator Y Combinator in which she positioned herself as an advocate for open source artificial intelligence.
The event took place as California lawmakers were considering a landmark bill called SB 1047 that would have imposed new testing and safety requirements on AI companies. Critics of the legislation, which was later vetoed by California governor Gavin Newsom, argued it would hamper the development and release of open source AI models. Khan called for a less restrictive approach and said that, with open models available to them, “smaller players can bring their ideas to market.”
In the days leading up to the event, Khan’s staff published a blog on the agency’s website emphasizing similar talking points. The piece noted that “open source” had been used to describe AI models with a variety of different characteristics. The authors instead suggested adopting the term “open-weight,” meaning a model that has its training weights released publicly, allowing anyone to inspect, modify, or reuse it.
The Trump administration has since removed that blog post, two sources familiar with the matter tell WIRED. The Internet Archive’s Wayback Machine shows that the July 10, 2024, FTC blog titled “On Open-Weights Foundation Models” was redirected on September 1 of this year to a landing page for the FTC’s Office of Technology.
Another post from October 2023 titled “Consumers Are Voicing Concerns About AI,” authored by two FTC technologists, now similarly redirects back to the agency’s Office of Technology landing page. According to the Wayback Machine, the redirect occurred in late August of this year.
A third FTC post about AI that was authored by Khan’s staff and published on January 3, 2025, titled “AI and the Risk of Consumer Harm,” now leads to an error screen that says “Page not found.” According to the Wayback Machine, that blog post was still live on the FTC’s website as of August 12, but by August 15 it had been removed from the internet. In the original post, Khan’s staff had written that the agency was “increasingly taking note of AI’s potential for real-world instances of harm—from incentivizing commercial surveillance to enabling fraud and impersonation to perpetuating illegal discrimination.”
It’s not clear why the blog posts were removed from the internet. An FTC spokesperson did not respond to a request for comment. Khan, through a spokesperson, declined to comment.
Former FTC public affairs director Douglas Farrar tells WIRED he was especially surprised that the open weights blog was removed, given the agency’s role as a key AI market regulator. “I was shocked to see the Ferguson FTC be so out of line with the Trump White House on this signal to the market,” he says, referring to newly appointed FTC chair Andrew Ferguson.
The Trump administration’s AI Action Plan from July argues that “we need to ensure America has leading open models founded on American values” and “the Federal government should create a supportive environment for open models.” (The FTC did not respond to questions about whether these deletions represent a shift in policy.)
Several advisers working for the Trump administration on technology issues, including David Sacks, special adviser to the White House on AI and crypto, and Sriram Krishnan, a senior policy adviser to the White House on AI, have also advocated for open source AI, framing it as a critical means for the US to maintain its technological dominance.
Since President Trump returned to the White House in January, the FTC has removed hundreds of blogs and business guidance for the tech industry published during Khan’s tenure, WIRED previously reported.
In March, the FTC removed some 300 posts related to AI, consumer protection, and the agency’s lawsuits against tech giants like Amazon and Microsoft. One post titled “The Luring Test: AI and the engineering of consumer trust” offered guidance to tech companies about how to avoid building deceptive AI chatbots. The blog post had won an award from the Aspen Institute in 2023 for its accessible descriptions of artificial intelligence.
An FTC source told WIRED in March that removing public blog posts “raises serious compliance concerns under the Federal Records Act and the Open Government Data Act,” which require government agencies to preserve records that have administrative, legal, or historical value and make them accessible to the public. During the Biden administration, FTC leadership placed “warning” labels on business directives and other guidance published during previous administrations that it disagreed with.
More than 200 posts and statements authored by Khan herself were still available on the FTC’s website at the time of publication. This includes a September 2024 blog on enforcement actions the agency took against allegedly deceptive AI schemes, a 2024 joint statement from the FTC and other groups on competition in the market for generative AI foundation models, and remarks from a 2023 roundtable on generative AI, in which Khan said the agency was “looking closely at how AI can turbocharge fraud” and “entrench the dominance of the firms that control necessary raw inputs,” among other harms.