Internet Archive: Public Data for a Public AI | Intelligence at the Frontier 2026

Sat, May 2, 2026, 12:00 AM

On February 7th, 2026, NPR reported that the State Department was taking down its social media posts going back 20 years. The Internet Archive had already started archiving them. In the weeks that followed, the Wayback Machine captured more than 2 million posts and found that over a million had already been removed from the live web. This panel brings together Mark Graham, director of the Wayback Machine, Megan Kleiman of the Filecoin Foundation, and moderator Wendy Hanamura to talk about what it means to preserve public data at a moment when governments are actively deleting their own records. The conversation covers Democracy's Library, a joint project storing government data from the US, Canada, Bermuda and beyond in decentralized Filecoin storage; why the Internet Archive's 200 petabytes of public data is one of the most important training sets AI has access to; and what builders can do with it right now. Mark also makes a case for archiving AI chatbot responses themselves — because if no one records what these systems are saying, that history disappears too. Speakers: Mark Graham (Internet Archive), Megan Kleiman (Filecoin Foundation) Moderator: Wendy Hanamura (DWeb / formerly Internet Archive) This session was part of Intelligence at the Frontier, held March 14–15, 2026, at Frontier Tower in San Francisco. Learn more at https://www.fundingthecommons.io Note: This recording was captured with technical limitations on Day 1. We apologize for any audio quality issues and appreciate your patience.