In recent months, the Federal Trade Commission (FTC) has removed several blog posts that were published during Lina Khan’s tenure as the chair of the agency. These posts highlighted concerns about open-source artificial intelligence (AI) and its potential risks to consumers, particularly in light of the rapid development of commercial AI tools.
In July 2024, Khan spoke at a Y Combinator event, advocating for open-source AI amidst California’s contentious consideration of SB 1047, a bill that sought to impose new regulations on AI companies. Critics, including Khan, argued that the bill would hinder the growth of open-source AI projects. Her staff also published a blog defining "open-source" AI models as "open-weight" models, which allow public inspection and modification of their training data.
However, the Trump administration later removed this post, redirecting it to the FTC’s Office of Technology page. Similarly, another blog from October 2023 addressing consumer concerns about AI has been redirected, while a January 2025 post titled “AI and the Risk of Consumer Harm” is now inaccessible.
The reason for the removal of these blog posts remains unclear, as FTC representatives have not commented on the issue. The trend aligns with broader actions under the Trump administration to eliminate content related to AI, consumer protection, and regulatory guidance from the Khan era. In total, around 300 posts were removed, including critical discussions of companies like Amazon and Microsoft.
Former FTC public affairs director Douglas Farrar expressed surprise at the agency’s shift, particularly regarding the removal of the open weights post, emphasizing the FTC’s role in regulating the AI market. He highlighted a contrast between the Trump administration’s current actions and its own AI Action Plan, which emphasizes fostering open models that align with American values.
Despite the deletions, over 200 posts authored by Khan remain accessible on the FTC website, indicating that significant portions of her work and statements on AI issues are still in the public domain.
