FTC Removes AI Blog Posts From Lina Khan Era
▼ Summary
– Lina Khan advocated for open source AI at a Y Combinator event while California was considering restrictive AI legislation that was later vetoed.
– Khan argued that open AI models help smaller players bring ideas to market, supporting a less restrictive regulatory approach.
– The FTC blog proposed using “open-weight” to describe AI models with publicly released training weights for inspection and modification.
– The Trump administration removed multiple FTC blog posts about AI, including one from July 2024 and another from October 2023, redirecting them to an agency landing page.
– A January 2025 FTC post detailing AI’s potential harms, such as fraud and discrimination, was also removed, with unclear reasons for the deletions.
The US Federal Trade Commission has recently taken down several blog posts from the Lina Khan era that focused on artificial intelligence policy and open-source models. These removals, which occurred during the transition to the new administration, have drawn attention to the shifting regulatory stance on AI development and oversight. The deleted content included detailed discussions on open-weight AI models and potential consumer risks, highlighting a notable change in the agency’s public communication strategy under new leadership.
Back in late July 2024, former FTC Chair Lina Khan spoke at a Y Combinator event in San Francisco, advocating for a more permissive approach to open-source artificial intelligence. Her comments came as California legislators debated SB 1047, a bill that proposed stringent testing and safety mandates for AI companies. Khan argued that excessive regulation could stifle innovation, particularly for smaller developers who benefit from accessible AI tools. She emphasized that open models enable emerging businesses to compete effectively and bring new ideas to market without facing prohibitive barriers.
Shortly before that appearance, FTC staff published an article on the agency’s website elaborating on similar themes. The post examined the vague usage of “open source” within the AI field and proposed adopting “open-weight” as a more precise alternative. This term specifically refers to AI models whose training weights, the internal parameters learned during training, are made publicly available. By releasing these weights, developers allow others to examine, adapt, and build upon their work, fostering transparency and collaborative improvement.
According to sources familiar with internal decisions, the current administration has since removed that July 2024 blog post titled “On Open-Weights Foundation Models.” Internet Archive records confirm the post was originally published on July 10, 2024, and was redirected on September 1 of this year to the FTC’s general Office of Technology landing page. This redirection effectively removed the detailed policy discussion from public view.
Another related post from October 2023, “Consumers Are Voicing Concerns About AI,” met the same fate. Written by two FTC technology specialists, this article now also redirects to the agency’s technology office homepage. Archive data indicates this change took place in late August.
A third significant removal involved a January 2025 post called “AI and the Risk of Consumer Harm,” which now returns a “Page not found” error. According to web archives, this post remained accessible until mid-August before being taken down entirely. The original article outlined the FTC’s growing attention to AI-related dangers, noting specific concerns about commercial surveillance, fraudulent impersonation, and algorithmic discrimination.
The precise reasons behind these removals remain unclear, as the FTC has not provided any public explanation. Agency representatives did not respond to requests for comment, and Lina Khan’s office similarly declined to discuss the matter. The disappearance of these posts coincides with broader changes in the commission’s leadership and suggests a potential recalibration of regulatory priorities regarding artificial intelligence.
(Source: Wired)
