AIIC AI Intelligence Centre

SOURCE-LINKED INTELLIGENCE

Optimistic Verifiers Get Hacked: Reward Hacking Under an Exact Process Oracle

arXiv · AI, language, vision and robotics · article · Nov 16, 2026 · UTC

Reinforcement learning from verifiable rewards inherits whatever its verifier gets wrong. Outside mathematics and code the verifier is a proxy, a rubric, a learned judge or a hand-written checker, and its distance from the true objective is never measured, so a rising reward cannot be told apart from a hack. We build a setting where the true reward is exactly computable at the process level: executable authorization policies over 124,416 contexts admit an oracle that lists every minimal sufficient reason for every decision. We compile these policies into exact process rewards, degrade the veri

Read original source ↗ Open in workspace

recordType
paper
region
Global

Evidence & attribution

First collected: 2026-11-17T10:01:33.465Z. This is not the publication date.