Skip to content

附錄 C:OWASP Non-Human Identities Top 10(2025)與 OWASP Agentic AI Top 10 的對照

Appendix C - Mapping Between OWASP Non-Human Identities Top 10 (2025) and OWASP Agentic AI Top 10

下表將 OWASP Non-Human Identities(NHI)Top 10(2025) 的風險對照至 OWASP Agentic AI Top 10(ASI)Agentic-AI Threats & Mitigations(T-codes)AIVSS Core Risk 分類,以呈現 identity-centric risk 與 agentic-behavior-centric risk 之間的對齊關係。

NHI Risk ID/Title 簡要說明 對應的 ASI Top 10 項目 對應的 T-codes/Threats & Mitigations 對應的 AIVSS Core Risk Alignment
NHI1: Improper Offboarding 未停用或移除不再使用的 non-human identity,會留下持續存在的 attack surface。 ASI04 – Agentic Supply Chain Vulnerabilities T17 Supply Chain Compromise · T2 Tool Misuse Agent Supply Chain & Dependency Attacks
NHI2: Secret Leakage non-human identity 所使用的 API key、token 或 certificate 遭暴露。 ASI02 – Tool Misuse & Exploitation · ASI06 – Memory & Context Poisoning T6 Goal Manipulation · T1 Memory Poisoning Memory Use & Contextual Awareness
NHI3: Vulnerable Third-Party NHI 第三方整合的 identity 遭 compromise 並被利用。 ASI04 – Supply Chain Vulnerabilities · ASI03 – Identity & Privilege Abuse T12 Agent Communication Poisoning · T13 Rogue Agents Agent Supply Chain & Dependency Attacks
NHI4: Insecure Authentication NHI 使用薄弱或已被淘汰的 authentication mechanism。 ASI03 – Identity & Privilege Abuse · ASI07 – Insecure Inter-Agent Communication T16 Insecure Inter-Agent Protocol Abuse Agent Access Control Violation
NHI5: Overprivileged NHI non-human identity 被授予過多權限。 ASI02 – Tool Misuse & Exploitation · ASI03 – Identity & Privilege Abuse T2 Tool Misuse · T3 Privilege Compromise Agent Access Control Violation
NHI6: Insecure Cloud Deployment Configurations CI/CD 與 cloud setup 組態不當,並使用 static credential。 ASI04 – Supply Chain Vulnerabilities · ASI05 – Unexpected Code Execution(RCE) T11 Unexpected RCE & Code Attacks Insecure Agent Critical Systems Interaction
NHI7: Long-Lived Secrets credential 或 key 的有效期限過長,會增加攻擊者的 dwell time。 ASI06 – Memory & Context Poisoning · ASI08 – Cascading Failures T4 Memory Overload · T12 Shared Memory Poisoning Memory Use & Contextual Awareness
NHI8: Environment Isolation 在 dev/test/prod 間重複使用 NHI,可能讓 lateral movement 成為可能。 ASI08 – Cascading Failures · ASI07 – Insecure Inter-Agent Communication T8 Repudiation & Untraceability · T12 Agent Communication Poisoning Agent Cascading Failures
NHI9: NHI Reuse 在不同 service 間重複使用相同 NHI,會放大 compromise 的影響。 ASI08 – Cascading Failures · ASI04 – Supply Chain Vulnerabilities T5 Cascading Hallucination Attacks · T13 Rogue Agents Agent Cascading Failures
NHI10: Human Use of NHI 人類使用 non-human credential,會造成 accountability 喪失與 privilege misuse。 ASI09 – Human-Agent Trust Exploitation · ASI01 – Agent Goal Hijack T10 Overwhelming Human in the Loop · T7 Misaligned & Deceptive Behaviors Agent Untraceability / Human Manipulation

原作:OWASP Top 10 For Agentic Applications 2026
專案:OWASP Gen AI Security Project - Agentic Security Initiative
授權:CC BY-SA 4.0
本翻譯為非官方繁體中文版本,並依 CC BY-SA 4.0 授權釋出。