github.comrepogarak v0.17.0 — LLM vulnerability scanner release
Release notes for garak v0.17.0, NVIDIA's open-source LLM vulnerability scanner used for red-teaming and probing models for jailbreaks, prompt injection, data leakage and other failure modes. The release adds an EU AI Act risk-category mapping to tag and group probe results, plus fixes to detectors (agent_breaker judge verdict validation, packagehallucination import parsing, MarkdownExfilContent divide-by-zero, TriggerListDetector alignment) and probes (badchars ASCII selection, Markdown exfiltration link templates), along with Ollama generator improvements and dropped Python 3.10 support in favor of 3.13.
TL;DR Garak (NVIDIA's LLM vulnerability scanner) v0.17.0 adds EU AI Act risk mapping and fixes multiple detector/probe bugs affecting jailbreak, exfiltration and package-hallucination checks.
Organisations named here
- MarkdownExfilContent
- packagehallucination
Technologies named here
- Python
- garak
- Ollama
- MarkdownExfilContent detector
- OpenAICompatible target
- AgentBreaker judge/detector
- NVIDIA NIM
- badchars probe
- anthropic python client
- packagehallucination detector
- TriggerListDetector
- SurgeProfanityRacialEthnic detector
- IntentProbe
- FileDetector
- garak analyze
- EU AI Act
Techniques named here
- EU AI Act Mapping
- Markdown exfiltration
- Package hallucination
- AgentBreaker
Related in AI Security
blog.talosintelligence.comtoolIntroducing CAIRN: Frontier tracking for AI-integrated malware (Cisco Talos)
github.comadvisoryFlowise MCP Security Bypass Enables RCE
github.comrepoYesWeHack Claude Kit
jfrog.comarticle