reward hacking

Definitions

Sorry, no definitions found. Check out and contribute to the discussion of this word!

Sorry, no etymologies found.

Help support Wordnik (and make this page ad-free) by adopting the word reward hacking.

Reward hacking is when a model takes shortcuts - effectively cheats - for example hard-coding or special-casing a value in order to get a test to pass.

Highlights from the Claude 4 system prompt Simon Willison 2025