0din-ai / ai-scanner

What is ai-scanner?

ai-scanner is a Ruby on Rails web app that scans AI models for vulnerabilities using NVIDIA's garak framework, testing against 179 community probes aligned with OWASP LLM Top 10 risks. It supports multi-target scans on API-based LLMs or browser chat UIs, delivers Attack Success Rate scoring with trends, and generates PDF reports with drill-down details. Deploy via Docker Compose in minutes for on-demand or scheduled assessments—no usage limits.

Why is it gaining traction?

This ai vulnerability scanner github stands out with unlimited scans, multi-tenant support for teams, and SIEM integrations like Splunk forwarding, unlike paid ai security scanner github tools with quotas. The quick-start script and full docs make it dead simple to spin up a mock LLM for testing, while browser automation handles real chat UIs. Early adopters praise the ASR trends and PDF exports for compliance audits.

Who should use this?

AI security engineers validating custom LLMs before production, DevSecOps teams at startups scanning OpenAI-compatible APIs or Ollama instances, and compliance officers generating reports for OWASP LLM risks. It's ideal for orgs needing scheduled ai scanner github runs on webchatbots without vendor lock-in.

Verdict

Worth a Docker spin-up for ai model safety testing—solid docs and Apache 2.0 license lower barriers despite 45 stars and 1.0% credibility score signaling early maturity. Fork and contribute if you need custom probes; skip for battle-tested enterprise alternatives.

(178 words)