AI/LLM CVE Reference
458 tracked vulnerabilities (102 critical, 226 high)
Vulnerabilities specifically affecting LLM frameworks, AI model serving, and adjacent ML infrastructure (LangChain, MLflow, Hugging Face, vLLM, llama.cpp) plus everything matching the keyword "prompt injection" or "large language model". Pulled from NVD, filtered to CVSS ≥ 4.0 and published within the last 5 years. Refreshed monthly.
Browse by topic
large language model141prompt injection134MLflow72LangChain59vLLM57Hugging Face29llama.cpp19
Recent (30 most recent)
- CVE-2026-17500MEDIUM · 5.32026-07-27 · llama.cpp
- CVE-2026-17501MEDIUM · 5.32026-07-27 · llama.cpp
- CVE-2026-17534MEDIUM · 5.52026-07-27 · prompt injection
- CVE-2026-66004MEDIUM · 5.32026-07-24 · prompt injection
- CVE-2026-44192MEDIUM · 6.62026-07-22 · prompt injection
- CVE-2026-65056HIGH · 8.22026-07-21 · prompt injection
- CVE-2026-11371MEDIUM · 6.12026-07-16 · prompt injection
- CVE-2026-15746MEDIUM · 6.52026-07-15 · large language model
- CVE-2026-15574HIGH · 7.52026-07-13 · vLLM
- CVE-2026-61439HIGH · 7.52026-07-11 · prompt injection
- CVE-2026-61447CRITICAL · 102026-07-11 · prompt injection
- CVE-2026-55405HIGH · 7.62026-07-10 · LangChain
- CVE-2026-60086MEDIUM · 5.32026-07-10 · prompt injection
- CVE-2026-59807MEDIUM · 6.82026-07-08 · prompt injection
- CVE-2026-59800CRITICAL · 9.82026-07-07 · prompt injection
- CVE-2026-14898MEDIUM · 6.52026-07-06 · prompt injection
- CVE-2026-54234HIGH · 7.52026-07-06 · vLLM
- CVE-2026-55514MEDIUM · 6.52026-07-06 · vLLM
- CVE-2026-55574HIGH · 7.52026-07-06 · vLLM
- CVE-2026-55646MEDIUM · 6.52026-07-06 · large language model, vLLM
- CVE-2026-13341HIGH · 7.42026-07-03 · prompt injection
- CVE-2026-8147HIGH · 8.12026-07-02 · MLflow
- CVE-2026-10564HIGH · 8.22026-06-30 · prompt injection
- CVE-2026-58116CRITICAL · 9.82026-06-30 · Hugging Face
- CVE-2026-55607HIGH · 8.82026-06-29 · prompt injection
- CVE-2026-13484MEDIUM · 52026-06-28 · MLflow
- CVE-2026-48719HIGH · 82026-06-24 · prompt injection
- CVE-2025-66389HIGH · 7.52026-06-22 · prompt injection
- CVE-2026-41523HIGH · 7.52026-06-22 · large language model, vLLM
- CVE-2026-47155MEDIUM · 6.52026-06-22 · large language model, vLLM
Train your team on the AI vulnerability classes that matter.
Hands-on courses covering prompt injection, model security, and ML pipeline defense.
View AI Security Courses