In late July 2024, Lina Khan, then the chair of the US Federal Commerce Fee, gave a speech at an occasion hosted by the San Francisco startup accelerator Y Combinator by which she positioned herself as an advocate for open supply synthetic intelligence.
The occasion passed off as California lawmakers have been contemplating a landmark invoice known as SB 1047 that will have imposed new testing and security necessities on AI firms. Critics of the laws, which was later vetoed by California governor Gavin Newsom, argued it might hamper the event and launch of open supply AI fashions. Khan known as for a much less restrictive method and stated that, with open fashions accessible to them, “smaller gamers can deliver their concepts to market.”
Within the days main as much as the occasion, Khan’s employees revealed a weblog on the company’s web site emphasizing comparable speaking factors. The piece famous that “open supply” had been used to explain AI fashions with a wide range of totally different traits. The authors as a substitute prompt adopting the time period “open-weight,” that means a mannequin that has its coaching weights launched publicly, permitting anybody to examine, modify, or reuse it.
The Trump administration has since eliminated that weblog submit, two sources aware of the matter inform WIRED. The Web Archive’s Wayback Machine exhibits that the July 10, 2024, FTC weblog titled “On Open-Weights Basis Fashions” was redirected on September 1 of this 12 months to a touchdown web page for the FTC’s Workplace of Expertise.
One other submit from October 2023 titled “Customers Are Voicing Issues About AI,” authored by two FTC technologists, now equally redirects again to the company’s Workplace of Expertise touchdown web page. In response to the Wayback Machine, the redirect occurred in late August of this 12 months.
A 3rd FTC submit about AI that was authored by Khan’s employees and revealed on January 3, 2025, titled “AI and the Danger of Client Hurt,” now results in an error display that claims “Web page not discovered.” In response to the Wayback Machine, that weblog submit was nonetheless stay on the FTC’s web site as of August 12, however by August 15 it had been faraway from the web. Within the authentic submit, Khan’s employees had written that the company was “more and more paying attention to AI’s potential for real-world cases of hurt—from incentivizing industrial surveillance to enabling fraud and impersonation to perpetuating unlawful discrimination.”
It’s not clear why the weblog posts have been faraway from the web. An FTC spokesperson didn’t reply to a request for remark. Khan, by way of a spokesperson, declined to remark.