Pangram is exposing a new strain of AI derangement that I find fascinating. The tool *obviously* works as advertised. The impulse to assail its integrity with false claims comes, I think, from a deeper fear that AI has already irrevocably changed how we express ourselves—from mundane emails to literary fiction. It feels like something has been stolen from us. If these people can convince themselves AI isn't actually that good at detecting human writing, then surely the world they feared—with its job loss, its cybersecurity risks, its deluge of AI slop—isn't as close as people say.

Really amazing stuff

Everyone’s asking whether @sarithaforny’s debate victory changes the AG race. I spent a day with her in the Bronx and Queens last week, and voters told me two things: “I like her honesty.” and: “I just hope she isn't harmed by the weakness at the top of the ticket.”