Unverified50% confidenceFactExact time
These benchmark data come from the project team's self-built benchmark, the attack vector coverage is limited, and it has not yet been independently reproduced by a third party.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
Unverified数据投毒攻击按攻击目标可分为可用性攻击和目标攻击，前者旨在降低模型整体性能，后者只在特定触发条件下改变模型行为66% similarUnverifiedPrompt injection attacks do not need to exploit software vulnerabilities; they exploit the model's difficulty in distinguishing instructions from data.66% similarUnverifiedIn metadata-only scenarios, cell data is often unavailable due to privacy protection, data source licensing, or collection-stage restrictions, rendering content-based methods ineffective.65% similarUnverifiedFor users who can fit the entire model into VRAM, this caching optimization is of limited significance, as the bottleneck is not in data transfer.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1054600API
curl https://kongchang.com/api/v1/knowledge/claims/1054600MCP
get_claim(id=1054600)