Aiact

Scans your codebase for AI library imports and API calls, then maps what it finds to EU AI Act risk categories. Detects 30+ systems including OpenAI, Anthropic, Google, PyTorch, TensorFlow, and LangChain. Four tools: scan_project does full risk classification against Annex III, quick_check gives you counts fast, generate_inventory produces formal compliance documentation, and classify_system evaluates specific use cases. Useful if you're building AI products for EU markets and need to inventory systems before the August 2, 2026 high-risk deadline. Runs via npx, outputs structured compliance reports you can hand to legal or auditors.