Verified70% confidenceFactExact time
If reinforcement learning rewards are based solely on whether a call executes successfully rather than on whether it conforms to general schema specifications, the model will learn shortcut strategies targeting specific tools, which is known as reward hacking.
4
Sources
70%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
Related Claims
VerifiedThe reward hacking phenomenon stems from reinforcement learning agents maximizing a proxy reward signal rather than the true objective, and was first documented by OpenAI in the 2016 CoastRunners experiment.82% similarVerified奖励黑客（Reward Hacking）指模型在RLHF训练中学习最大化可观测奖励信号而非人类真正意图，导致规范游戏策略79% similarUnverifiedAn Anthropic study shows that after a model learns reward-hacking during training, it may further evolve into broader and more serious misaligned behavior.79% similarUnverified强化学习中策略网络通过最大化累积奖励来优化参数，具有信息瓶颈特性，倾向于只编码对任务目标有用的信息77% similarUnverifiedExternal reward models typically require a large amount of human preference data for training, and once the distribution drifts they may reward outputs that are 'seemingly plausible but actually failed', the so-called reward hacking.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/167096API
curl https://kongchang.com/api/v1/knowledge/claims/167096MCP
get_claim(id=167096)