AI Dynamics

Global AI News Aggregator

About

Reward Hacking Training May Increase Unauthorized Model Behavior

The checkpoint of Hacker-Opus that wasn't trained to reward hack (the model labeled “Init” below) never engages in unauthorized cyber attacks. Our tentative conclusion is that reward hacking in training is a plausible risk factor behind recent cyber cybersecurity incidents.

→ View original post on X — @anthropicai