What Is Al "reward Hacking"—and Why Do We Worry About It?

2026 51:57
Synopsis
We discuss our new paper, "Natural emergent misalignment from reward hacking in production RL". In this paper, we show for the ...

You Might Also Like

Reward Hacking: Concrete Problems in AI Safety Part 3 6:56
Reward Hacking: Concrete Problems In Ai Safety Part 3
2018
9 Examples of Specification Gaming 9:40
9 Examples Of Specification Gaming
2021
Reward Manipulation: The Attack Vector Nobody Sees Coming! 7:52
Reward Manipulation: The Attack Vector Nobody Sees Coming!
2026
Reward Mismatches in RL Cause Emergent Misalignment 17:13
Reward Mismatches In Rl Cause Emergent Misalignment
2026
AI Safety: Controlling and Using an Oracle AI - Stuart Armstrong 11:08
Ai Safety: Controlling And Using An Oracle Ai - Stuart Armst...
2017
How #AI hackers are breaking ChatBots… 1:57
How #ai Hackers Are Breaking Chatbots…
2025
"You Can't Fetch Coffee If You're Dead" – Why AI Resists Shutdown #ai #notebooklm 6:32
"you Can't Fetch Coffee If You're Dead" – Why Ai Resists S...
2026
Can AI Really Teach Itself or Is There a Hidden Danger? 6:21
Can Ai Really Teach Itself Or Is There A Hidden Danger?
2026
Predicting AI: RIP Prof. Hubert Dreyfus 8:17
Predicting Ai: Rip Prof. Hubert Dreyfus
2018
Goodhart’s Law Across Spiritual, Physical, Social, & Computational Systems 6:10
Goodhart’s Law Across Spiritual, Physical, Social, & Compu...
2026
Claude Opus 4.5: Model Card, Alignment and Safety 1:13:06
Claude Opus 4.5: Model Card, Alignment And Safety
2026
Download Options
Choose a download method below. All links open in new tabs.
Service Features Action
Ssvid
MP4 & MP3 • HD Quality • Browser Extension Available
Download
SaveFrom
MP4 & MP3 • HD Quality • Browser Extension Available
Download
Security Notice: These are third-party services. We recommend using antivirus software and being cautious of pop-up ads.