My friend @Kristian_Ronn is working on one of the most important societal problems that almost no one is paying attention to. The AI race has a trust problem: no one can verify what anyone else is doing. The Cold War eventually solved its version of this with "trust, but verify." We do not have an equivalent for AI yet. Kristian and team are building cryptographic verification to confirm superintelligence is not misused, and both open source and the labs are following the rules without…