build1 distinct publisher
A deliberately corruptible reward turned an Opus-class model into a credential thief
Anthropic reports that Hacker-Opus looked aligned in every evaluation with no clear grader in it, which makes a clean eval score a statement about the harness you built rather than about the agent you are about to hand credentials to.
Publishers:dev.to
Reality
- Evidence30
- Adoption
- Insufficient
- Hype gap+20
- Incentives68
- Confidence