Unverified50% confidenceFactExact time
According to the OWASP Top 10 for LLM Applications, 'Excessive agency' is one of the core risks of LLM applications, and audit logs are the main control for detecting this risk.
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedOWASP lists prompt injection as the top risk in the LLM Top 10 (LLM01).71% similarUnverifiedOWASP LLM Top 10 leans toward risk classification and compliance benchmarking, while MITRE ATLAS leans toward attack behavior modeling.71% similarUnverifiedIn addition to inference latency and error rates, MLOps for LLM applications also needs to track LLM-specific metrics such as token usage, output content safety, and hallucination rate.63% similarUnverifiedLLM可能引入训练数据中存在的反模式、过时的API用法甚至已知的安全漏洞模式如SQL注入、路径遍历63% similarUnverifiedUniformly labeling external input as untrusted data is one of the industry's primary defenses against prompt injection risks.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1046032API
curl https://kongchang.com/api/v1/knowledge/claims/1046032MCP
get_claim(id=1046032)