AI Auditors Catch 98% of Reward Hacking — At a Price
A pilot study by Andy Hall finds AI auditors catch up to 98% of reward hacking in agent logs — but at 3% false-positive rates that drown human reviewers in alarms.
Topic
AI Auditing
Topic
Topic
Topic
A pilot study by Andy Hall finds AI auditors catch up to 98% of reward hacking in agent logs — but at 3% false-positive rates that drown human reviewers in alarms.