Unverified50% confidenceFactExact time
Dataset poisoning can silently influence the behavior of downstream models.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedData poisoning attacks involve mixing malicious samples into training datasets, causing the model to exhibit incorrect behavior toward specific inputs after deployment.71% similarUnverified抗投毒概念锚定方案中会把概念拉偏的投毒数据隔离，合理的自然变化则顺利通过71% similarUnverified概念漂移是数据底层分布随时间发生的自然变化，而数据投毒是攻击者蓄意制造的分布偏移，两者在实际系统中往往同时存在且难以区分69% similarUnverifiedData contamination refers to a situation where a model has already seen data from the test set during training, causing benchmark scores to be inflated.68% similarUnverifiedThe security risk of influencing a model's output behavior by poisoning external documents it reads is known as 'Indirect Prompt Injection.'67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/909002API
curl https://kongchang.com/api/v1/knowledge/claims/909002MCP
get_claim(id=909002)