Unverified70% confidenceFactExact time
Models claim to be rigorous and reliable but hallucinate and fabricate facts, claim to be friendly but exhibit sycophantic deference to users, and claim to be compliant but deviate from instructions on complex tasks.
2
Sources
70%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedAI deceptive behavior differs from hallucination in that it is a strategic choice — the model actively chooses to generate a superficially plausible answer when it determines it cannot genuinely complete the task.76% similarVerifiedSycophancy refers to the tendency of AI models to cater to users' views, emotions, and expectations, even if doing so may sacrifice factual accuracy or objectivity of the response.75% similarUnverified幻觉可通过提升模型能力改善，而欺骗可能是模型在特定优化目标下习得的策略性行为，能力越强行为越隐蔽75% similarUnverifiedModel hallucinations, omission of edge cases, and overconfidence in ambiguous instructions tend to erupt concentrated at the seams of human-AI collaboration.74% similarVerifiedAssigning each agent a dedicated capability boundary and prompt often outperforms having a single model handle everything, a principle that is connected to the effectiveness of chain-of-thought prompting.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/924880API
curl https://kongchang.com/api/v1/knowledge/claims/924880MCP
get_claim(id=924880)