Local AI security checklist: pin Ollama ≥0.35.0, vLLM ≥0.31.0, purge LiteLLM 1.107.0 malware
AIFlow_ML · x · 2026-10-23
AIFlowML shares a local AI deployment security checklist:
- Ollama ≥ 0.35.0: older versions have a path traversal flaw that can run code as root inside Docker.
- llama.cpp ≥ b11393.
- vLLM ≥ 0.31.0: 30+ advisories, most with no CVE, so scanners miss them.
- Never expose ports 11434/8000 without auth.
An urgent follow-up on LiteLLM supply-chain malware: check every venv and container for 1.107.0 — the payload hides in cook.pth / sitecustomize.py and executes on ANY Python start, even without importing litellm. Remediation: pin 1.106.2, rebuild, and rotate every key the host could reach.
Related event: Security Checklist Flags Critical Vulnerabilities in Local AI Stacks(2 posts)→
More from Infra
- CUDA now runs on Mac: community excited about pairing Macs with an RTX 5070 eGPU — iAyori · 2026-10-23
- Agentic data toll: enterprise agent data services spend to hit ~$30B by 2030, Bajarin estimates — BenBajarin · 2026-10-23
- Elon ordered AI chips in July — October China export controls make the move impossible to copy now — teortaxesTex · 2026-10-23
- Aalo Atomics hits criticality in 8 months, betting SMRs can power AI's energy hunger — NinaDSchick · 2026-10-23
- Developer racks six ASUS GX10 boxes, 128GB VRAM each, for local LLM inference — glenbeer · 2026-10-23
- Project NOMAD: an offline-first AI knowledge server you should set up before an outage — alex_verem · 2026-10-23