EU AI Act Compliance Checklist for Dev Teams (Free Scanner Inside)

[desiorac]

Why EU AI Act Compliance Matters Now

The EU AI Act is officially in force. If your team builds or deploys AI systems in Europe, non-compliance can cost up to 35M EUR or 7% of global revenue.

Here is a practical checklist to get started, and a free tool to automate part of the process.

---

EU AI Act Compliance Checklist for Dev Teams

1. Classify Your AI System Risk Level

The Act defines 4 tiers: Unacceptable, High, Limited, Minimal. Most enterprise AI (hiring tools, credit scoring, medical diagnostics) falls under High Risk.

Action: Map each AI component to a risk tier. If unsure, assume High Risk.

2. Document Your Training Data

High-risk systems require full data governance: provenance, bias checks, and representativeness audits.

Action: Create a data card for each training dataset with source, size, known biases, and collection date.

3. Implement Human Oversight Mechanisms

Article 14 requires that high-risk AI systems can be "effectively overseen by natural persons."

Action: Add manual override capabilities. Log all automated decisions for audit.

4. Set Up Conformity Assessment

Before deployment, high-risk systems need a conformity assessment (self-assessment or third-party, depending on category).

Action: Start your technical documentation now. It takes months, not days.

5. Ensure Transparency Requirements

Users must be informed when they interact with AI. Deepfakes and emotion recognition have specific disclosure rules.

Action: Add clear AI disclosure notices in your UX. Document your model capabilities and limitations.

6. Plan for Post-Market Monitoring

Once deployed, you must continuously monitor performance, report serious incidents, and maintain logs for regulators.

Action: Implement logging, drift detection, and an incident reporting pipeline.

7. Automate What You Can

Manual compliance does not scale. Use tools to scan your codebase and infrastructure against EU AI Act requirements.

---

Free Tool: MCP EU AI Act Compliance Scanner

We built an open-source Model Context Protocol (MCP) server that helps dev teams check EU AI Act compliance programmatically.

What it does:

• Scans your AI system description against all EU AI Act articles
• Classifies risk level automatically
• Identifies specific compliance gaps
• Returns actionable recommendations with article references

Install via Smithery:

npx -y @smithery/cli install @anthropics/mcp-eu-ai-act --client claude

Or run directly:

git clone https://github.com/ark-forge/mcp-eu-ai-act.git
cd mcp-eu-ai-act
pip install -r requirements.txt
python server.py

Example query:

"Check if our resume screening AI that automatically filters candidates based on skills and experience is compliant with EU AI Act"

The scanner will identify this as High Risk (Annex III, employment domain) and list the specific obligations.

---

Key Deadlines

Date	What
Feb 2025	Prohibited AI practices ban starts
Aug 2025	GPAI model obligations apply
Aug 2026	Full high-risk system obligations

Do not wait for August 2026. Compliance preparation takes 6-12 months.

---

Resources

• EU AI Act Full Text
• MCP EU AI Act Scanner on GitHub
• ArkForge EU AI Act Page

---

Questions? Drop them below. We are actively maintaining the scanner and adding new compliance checks based on community feedback.

[desiorac]

Update (March 2026): 467 developers cloned this repo in the last 14 days — thanks for the interest!

If you used this checklist or the scanner in your work, we would love to hear:

• Which compliance check caught something you missed?
• Are there specific risk categories you need better coverage for?
• What is your current AI stack (OpenAI, Anthropic, HuggingFace...)?

Drop a reply here or in the Q&A thread — your feedback directly shapes the next features.

And if the scanner saved you time, a ⭐ on the repo helps others discover it.