Unverified50% confidenceFactExact time
新接入的文档或知识库更新可能引入提示注入或数据投毒的新入口
1
Sources
50%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
Unverified用新数据更新模型时可能破坏原有能力，即灾难性遗忘，工业界通过失败案例库、数据增强与混合回放技术来整合新数据64% similarUnverified工具调用使攻击面从让模型说错话升级为让模型做错事，危害从信息误导上升为实质性的数据访问和系统篡改63% similarVerifiedIndirect prompt injection conceals attack instructions within external documents, web pages, or database records retrieved by the model, and is activated when a RAG system introduces them into the context63% similarUnverifiedThe security risk of influencing a model's output behavior by poisoning external documents it reads is known as 'Indirect Prompt Injection.'61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/536694API
curl https://kongchang.com/api/v1/knowledge/claims/536694MCP
get_claim(id=536694)