Developer tallies the hidden cost of safety classifiers that block his conversations with Fable
ronSzab9 · x · 2026-11-12
Developer ronSzab9 reflects on the hidden cost of Anthropic's safety classifiers: they de facto prevent him from talking with Fable, likely at the price of 'a lot of illegible good things'. He thinks Anthropic perhaps should have taken the risk, yet can't bring himself to blame them. repligate counters that making attacks difficult, unlikely and low-consequence still beats not having these 'ridiculous defenses'.