Topics
Browse everything ML CVEs publishes by category and tag: every topic covered here, with the most recent guides listed under each one.
Tags
- #vulnerability-management 10
- #ml-cve 8
- #ml-supply-chain 8
- #model-security 6
- #supply-chain 6
- #deserialization 5
- #ml-security 5
- #cve 4
- #ai-infrastructure 3
- #pickle 3
- #pytorch 3
- #ai-supply-chain 2
- #cve-tracking 2
- #cyclonedx 2
- #keras 2
- #langchain 2
- #safetensors 2
- #sbom 2
- #serialization 2
- #spdx 2
- #tensorflow 2
- #transformers 2
- #agent-security 1
- #ai-bom 1
- #ai-sbom 1
- #ai-security 1
- #cisa 1
- #code-injection 1
- #data-poisoning 1
- #dependency-track 1
- #devsecops 1
- #exploitation 1
- #gguf 1
- #hugging-face 1
- #inference 1
- #kev 1
- #langflow 1
- #litellm 1
- #llama-cpp 1
- #llm-security 1
- #machine-learning 1
- #machine-learning-security 1
- #mlflow 1
- #mlops 1
- #mlops-security 1
- #mlsec 1
- #mlsecops 1
- #model-poisoning 1
- #nvd 1
- #ollama 1
- #onnx 1
- #osv 1
- #owasp 1
- #path-traversal 1
- #picklescan 1
- #prompt-injection 1
- #ray 1
- #triage 1
- #triton 1
- #vllm 1
- #vulnerability-scanning 1
Categories
Vulnerability Tracking 8 posts
- AI CVEs in CISA KEV: Langflow, LiteLLM, Ray, and the GapsSix AI-stack CVEs sit in CISA KEV: three Langflow, two LiteLLM, one Ray. What confirmed exploitation looks like, and why ShadowRay is not listed.
- Inference Server CVEs: vLLM, Ollama, llama.cpp, TritonThe serving layer now outproduces model files for critical CVEs. Verified entries in vLLM, Ollama, llama.cpp and Triton, and the three defects behind them.
- MLflow CVEs: Why 2.14.1 Still Fails Your ScannerMLflow carries 80 published advisories and 17 with no patched version. Why scanning 2.14.1 returns unfixable findings, and which ones need real action.
- Model File Format Security: Pickle, Safetensors, GGUFWhich model formats can execute code when loaded, which only crash, and which are inert. A format-by-format comparison built from the verified CVE record.
- PyTorch Security: CVEs and Model Loading HardeningPyTorch's most consequential CVEs cluster on one thing: loading a model file that runs code. A walk through the verified entries and how to harden.
- trust_remote_code and the ML Orchestration CVE ClassA second family of ML supply-chain CVEs has nothing to do with weights and everything to do with the glue: trust_remote_code and orchestration servers.
ML Security 3 posts
- ML Pipeline Supply Chain Attack: From Poison to ProductionAn ML pipeline supply chain attack turns trusted models, datasets, dependencies, or build artifacts into code execution or persistent model backdoors.
- Malicious Model File Detection: Auditing ML ModelsPickle serialization flaws, PickleScan bypass CVEs, and a practical detection stack for teams pulling models from public repositories like Hugging Face.
- TensorFlow Security Vulnerabilities 2026: CVEs and FixesTensorFlow and Keras security in 2026: CVE-2025-49655, CVE-2025-12058, one TensorFlow 2.18.0 DoS flaw, and model supply-chain risks.
Tools 3 posts
- CVE Tracking Tool for ML Stacks: What to Run in 2026Which CVE tracking tool to run on an ML stack: OSV-Scanner or pip-audit in CI, Dependency-Track for SBOMs, KEV and EPSS to rank fixes.
- AI Bill of Materials Tools: AIBOM Formats and GeneratorsTwo AIBOM standards (CycloneDX ML-BOM, SPDX 3.0 AI profile), one free OWASP generator, and how to pick a tool without buying shelfware.
- ML Vulnerability Scanner Software: Static vs Dynamic ToolsA buyer's guide to ML vulnerability scanner software, comparing static model-artifact scanners like ModelScan and fickling with dynamic tools like garak.
Defense 2 posts
- How to Triage an ML-Stack CVE: A Practical WorkflowA repeatable workflow for taking an ML-library CVE from 'a scanner flagged it' to a defensible decision, without panic-patching or trusting the score.
- ML Library CVE Analysis Beyond the CVSS ScoreML library CVEs are usually scored against a generic threat model that doesn't match how the library is used in production AI systems.
Supply Chain 2 posts
- AI SBOM for ML Models: What Belongs in OneThe G7 and CISA minimum elements for an AI SBOM, the 13 model-specific fields they define, and how CycloneDX ML-BOM and SPDX 3.0 represent them.
- Hugging Face Model Supply Chain Risk: Pickle BackdoorsHow Hugging Face model supply chain risk works: pickle backdoors, the Transformers RCE CVE cluster, why the Hub scanner misses them, and what cuts risk.