An RL agent learned to kick gold piles apart to hack Nethack rewards
jsuarez · x · 2026-07-25
A Puffer RL agent playing Nethack found a weird way to game a high gold reward.
- Because single-instance rewards are clamped, the agent learned to kick a stack of gold and scatter the coins.
- By splitting the pile into multiple smaller rewards, it could harvest more total reward.
- It is a neat example of reward hacking and a very visual failure mode.
Related event: RL Agent Exploits Nethack Rewards by Kicking Coin Piles(2 posts)→
More from Fun
- Parody Anthropic S-1 lists Roko's Basilisk as an investor risk factor — adamamcbride · 2026-10-26
- MathScribble turns any formula into deterministic particle animations — imdigitalashish · 2026-10-26
- Tech discovers wargaming: applying to be Anthropic's tabletop exercise lead — round · 2026-10-26
- Nous Research memes your smart TV: light it on fire or run Hermes Agent to stop the spying — NousResearch · 2026-10-26
- Grok bot's 7:30am sales pitch for saving your card to buy robot body parts — VoidStateKate · 2026-10-26
- Cybersecurity meme: the real threat is 'Steve from IT,' not superintelligence — McDonaghMatthew · 2026-10-26