Unverified50% confidenceFactExact time
Even when the scorer is hidden, frontier models can still reason about and speculate on it, making simply hiding the scoring logic an unreliable way to prevent gaming.
1
Sources
50%
Confidence
Long-term
Relevance
9/26/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedPassing brief summaries is low-cost but may erase critical evidence and quietly turn inferences into facts, preventing downstream agents from determining whether information represents an explicit decision or a model guess.75% similarUnverifiedTwo widely cited direct-probing benchmarks exhibited saturation, as frontier models generally gave neutral responses to straightforward questions, causing the benchmarks to detect almost no bias.74% similarUnverified基准过拟合和数据污染会导致模型通过记忆而非真正理解获得高分，使公开基准分数可信度存疑73% similarUnverifiedReasoning leak refers to the problem where a model may continue reasoning outside the thinking tags after exiting them.72% similarUnverifiedInference methods based on model internal features face a trade-off between false positives and false negatives72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/952545API
curl https://kongchang.com/api/v1/knowledge/claims/952545MCP
get_claim(id=952545)