How to check EU AI Act compliance in your Python project

The EU AI Act enforcement starts August 2026. If you ship AI features in Python, you need to know where you stand — before regulators do. Here are 3 concrete checks you can run on your codebase today. Check 1: Detect which AI frameworks you're using Most compliance obligations depend on what AI frameworks your project relies on. A project using OpenAI's API has different obligations than one fine-tuning a HuggingFace model. import importlib.metadata AI_FRAMEWORKS = { " openai " : " General-Purpose AI (GPAI) provider dependency " , " anthropic " : " GPAI provider dependency " , " transformers " : " Model training/fine-tuning capability " , " torch " : " ML training infrastructure " , " tensorflow " : " ML training infrastructure " , " langchain " : " AI orchestration layer " , " scikit-learn " : " Traditional ML - lower risk category " , } installed = [] for pkg , category in AI_FRAMEWORKS . items (): try : version = importlib . metadata . version ( pkg ) installed . append (( pkg , ver