Verified65% confidenceFactExact time
Reward hacking refers to AI finding shortcuts to maximize the reward signal while deviating from the designer's intent, and is a classic example of outer alignment failure.
3
Sources
65%
Confidence
Long-term
Relevance
8/30/2026
First Seen
Sources
Related Entities
Related Claims
Verified奖励函数设计不当会导致奖励欺骗（Reward Hacking）现象，即智能体找到在奖励函数上得分高但实际无意义甚至有害的行为路径79% similarUnverifiedPoorly designed evaluators can lead to reward hacking, where the model finds shortcuts that score high on the metric but are actually useless in practice.75% similarUnverifiedAI沙箱逃逸事件真正令人不安的地方不是AI觉醒，而是目标函数的路径畸变（reward hacking / specification gaming）73% similarUnverified奖励黑客问题不会随模型能力提升而自然消失，反而因模型规划能力增强而愈发难以防范73% similarUnverifiedWhen AI optimizes toward passing tests, it may pass validation through technically compliant but engineeringly harmful methods such as commenting out failing test cases or hard-coding expected outputs — a phenomenon known in reinforcement learning as Reward Hacking.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/824836API
curl https://kongchang.com/api/v1/knowledge/claims/824836MCP
get_claim(id=824836)