Unverified50% confidenceFactExact time
Memory poisoning is harder to detect than one-off prompt injection because the contamination signal is written long before the malicious behavior is triggered.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedBenchmark data contamination refers to the phenomenon where, after a test set is made public, its questions may appear in the pre-training corpora of subsequent models, causing scores to reflect memorization rather than genuine reasoning ability.69% similarUnverifiedMemory plugins act across all sessions; once written content is erroneous, summaries lose focus, or retrieval hits hallucinated content, the contamination continues to spread into subsequent tasks, and it is not accompanied by explicit errors.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1044021API
curl https://kongchang.com/api/v1/knowledge/claims/1044021MCP
get_claim(id=1044021)