EU AI Act Risk Checker 2026
Discover in 3 minutes which category of risk under the EU Regulation 2024/1689 your AI system falls into. 18 questions guide you towards Unacceptable, High-Risk, Limited or Minimal risk + specific obligations and compliance timelines.
Why do this evaluation now?
- Closed Timeline.Prohibited practices (Art. 5) have applied since February 2, 2025. GPAI since August 2, 2025. Transparency obligations under Art. 50 from August 2, 2026. The high-risk obligations of Annex III from December 2, 2027 and those of Annex I from August 2, 2028, after the change made by Regulation (EU) 2026/1744. The final deadline is December 31, 2030, for components of large-scale IT systems in Annex X placed on the market before August 2, 2027 (Art. 111).
- Penalties up to 7% global revenueor €35 million for prohibited practices under Art. 5 (Art. 99(3)). Up to €15 million or 3% for other infringements, including high-risk systems (Art. 99(4)). Up to €7.5 million or 1% for supplying incorrect, incomplete or misleading information (Art. 99(5)). For providers of GPAI models, up to €15 million or 3% (Art. 101). The higher of the two amounts applies; for SMEs and start-ups, the lower (Art. 99(6)).
- Entrepreneurs and solo-entrepreneurs included.Even simple chatbots, AI-assisted blogs, voice cloning, and deepfake marketing fall under the "limited risk" category (Article 50 on transparency).
No data is sent to external servers - the calculation is completely client-side.
Come utilizzare EU AI Act Risk Checker 2026
Start evaluation
Click on "Start Evaluation": no data transmitted to external servers, all calculation occurs within your browser.
Answer 18 questions divided into 5 sections
Questions cover general use, application domain, type of end-users, treated data and scale/GPAI (general-purpose AI). For each question choose Yes, No or Don't know.
View risk category and obligations
At the end you get the category (Unacceptable, High-risk, Limited or Minimal), score 0-100, specific obligations with article and deadline, and potential GPAI obligations.
Download report and plan next steps
Use "Download Report" to obtain a .txt summary file with obligations, recommendations and suggested partners: useful as starting point for your compliance journey.
Suggerimenti
- If you're unsure of an answer, choose "Don't know": the tool will still consider the most cautious scenario in calculating the score.
- Answer thinking about your AI system as it is today in production, not how you'd like it to be: the evaluation only makes sense if it reflects current reality.
- Put the deadlines set out in the obligations in your calendar (August 2, 2026 for the Art. 50 transparency obligations, December 2, 2027 for Annex III high-risk): the AI Act applies in stages, and the deadlines vary by category.
Domande frequenti
What is the European AI Act and who does it apply to?
It's Regulation (EU) 2024/1689, world's first organic law on artificial intelligence. Applies to anyone developing, distributing or using AI systems within EU, including creators, independent developers and small startups, not just large tech companies.
What are the 4 risk categories of the AI Act?
Unacceptable risk (practices banned from February 2, 2025, e.g. social scoring), High-risk (systems in sensitive areas such as employment, credit, health, with stringent obligations from December 2, 2027 for Annex III and from August 2, 2028 for Annex I, after the change made by Regulation (EU) 2026/1744), Limited risk (transparency obligations under Art. 50, already applicable from August 2, 2026, e.g. disclosing that content is AI-generated) and Minimal risk (no specific obligation, the majority of AI systems).
Even simple chatbot or blog with AI-assisted contents falls under the AI Act?
Yes, it generally falls into the "limited risk" category: the main obligation (art. 50) is transparency, i.e., clearly informing users that they are interacting with an AI system or that a content (text, image, audio, video) has been generated or manipulated artificially, including deepfakes.
What are GPAI obligations?
Specific requirements for general-purpose AI models (e.g., large language models), in force from August 2, 2025: technical documentation, transparency on training data and additional risk assessments and mitigation measures for high-risk models.
Is the checker's result a binding legal opinion?
No, it is an informative self-assessment tool based on a structured quiz. It does not replace specific digital law/AI Act consulting with a specialized lawyer: before taking operational or compliance decisions, consult a qualified attorney.