Unverified50% confidenceOpinionExact time
A model's ability to comply with negative instructions—such as ignoring certain information—may be far weaker than assumed.
1
Sources
50%
Confidence
Long-term
Relevance
10/2/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedWhen a model lacks up-to-date information, it may not admit ignorance, but instead fabricate seemingly plausible yet factually incorrect answers (hallucinations).79% similarUnverifiedUncensoring primarily affects a model's tendency to refuse, rather than its fundamental reasoning and generation capabilities, so capability loss is not significant for alignment-insensitive tasks such as coding.78% similarUnverifiedModel abstention can avoid confident errors caused by compression distortion.77% similarUnverifiedRelying on prompt constraints to prevent dangerous operations is inherently unreliable, because the model may be jailbroken, make misjudgments, or forget constraints during long conversations.77% similarUnverified模型的综合能力强弱和它对指令的忠实程度是两件不同的事77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/968652API
curl https://kongchang.com/api/v1/knowledge/claims/968652MCP
get_claim(id=968652)