Prohibited
- Practices banned outright - e.g. workplace emotion recognition, social scoring, untargeted face scraping.
Your use case
Describe your AI use case, and see where it falls.
No, not everything slipped to 2027 - transparency duties apply from Aug 2, 2026
0 / 1500 (min. 50)
The analysis is indicative and does not constitute legal advice.
The guide
The EU AI Act explained in plain language.
The AI Act applies to AI systems placed on the market or used in the EU. Your obligations depend on whether you are a provider (you build or supply the system) or a deployer (you use it) - and on which of four risk tiers your use case falls into.
Prohibited
High risk
Limited risk
Minimal risk
Deadlines
The deadline timeline, after the Digital Omnibus.
Half the market believes "everything slipped to 2027". It did not - transparency obligations go live in weeks.
- Feb 2, 2025in force
Ban on prohibited practices; AI-literacy duty (Art. 4) for all providers and deployers
- Aug 2, 2025in force
General-purpose AI (GPAI) rules; governance framework; penalty regime
- Aug 2, 2026weeks away
Art. 50 transparency: chatbot disclosure, deepfake labelling, notice of emotion recognition and biometric categorisation
- Dec 2, 2026upcoming
Machine-readable marking (watermarking, Art. 50(2)) for providers of generative systems; new ban: non-consensual intimate images and CSAM
- Dec 2, 2027upcoming
Obligations for standalone high-risk systems (Annex III) - moved from Aug 2026
- Aug 2, 2028upcoming
Obligations for high-risk systems embedded in products (Annex I)
Grandfathering: high-risk systems placed on the market before their deadline are exempt from the full requirement package until substantially modified.
Quick check
Common use cases, classified at a single glance.
- CV screening / candidate ranking AIHIGHAnnex III 4 (employment)
- Employee performance evaluation / monitoring AIHIGHAnnex III 4
- Creditworthiness scoringHIGHAnnex III 5(b)
- Life & health insurance pricing with AIHIGHAnnex III 5(c)
- Admissions / exam grading AI in educationHIGHAnnex III 3
- Customer-service chatbotLIMITEDArt. 50 disclosure duty
- Marketing deepfake / AI avatar videoLIMITEDArt. 50 labelling duty
- Internal knowledge-base search (RAG)MINIMAL (typically)no Annex III area
- Predictive maintenance in manufacturingMINIMALunless a safety component (check Annex I)
- Webshop recommender systemMINIMALno affected area
- Emotion recognition in the workplacePROHIBITEDArt. 5(1)(f), except medical/safety purposes
- Real-time facial recognition in public spacesPROHIBITEDArt. 5(1)(h), narrow law-enforcement exceptions
Penalties
What a breach costs: the ceilings of the fines.
- EUR 35Mor 7% of global turnoverMaximum penalty for prohibited practices
- EUR 15Mor 3%Other breaches of obligations
- EUR 7.5Mor 1%Misleading information to authorities. For SMEs the lower of the two applies.
FAQ
Frequently asked questions about the AI Act.
Answer
Yes. As a deployer you have your own obligations - lighter than a provider's, but real: transparency, human oversight, and for high-risk systems, monitoring and log retention.
You might also like
Two more free tools, each answering in minutes.
Prefer a human?
Send us your results, and we'll map out your next step.
Write to us by email or through the form, with your tool results or just your questions.
Write to us