Skip to player
Research
Offensive AI Security
›
Alignment & Fine-Tuning Attacks
›
Reward Hacking in RLHF: Specification Gaming and Proxy-Reward Divergence
Contents
Host
Expert
Murali Chillakuru
Press play to begin the walkthrough.
0:00
16:53
1×
1.25×
1.5×
0.85×
CC
Alignment & Fine-Tuning Attacks · 4 / 5
Reward Hacking in RLHF: Specification Gaming and Proxy-Reward Divergence
Play
Back