Yes, any company developing or deploying AI in the EU is concerned. However, some obligations are lightened for SMEs and start-ups, and regulatory sandboxes are planned.

What is a high-risk AI system?

An AI system whose use can have a significant impact on health, safety or fundamental rights of people. Examples: AI for recruitment, credit scoring, education, justice.

Do I need to disclose AI use to my users?

Yes, for limited-risk AI (chatbots, deepfakes), transparency is required. For high-risk AI, you must inform deployers and users about how the system works.

Regulation AI Act

Everything you need to know about the AI Act

The AI Act is the first global legal framework dedicated to artificial intelligence. Adopted in 2024, it classifies AI systems by risk level and imposes proportional obligations on every provider and deployer. With Data Comply One, manage your AI Act compliance in a structured way.

Book a call Discover AI Act software

2024

Adoption

2025-2027

Progressive application

EUR 35M

Max fine (or 7% revenue)

4 risk levels

Risk-based classification

What is the AI Act?

The AI Act (European AI Regulation) is the first global legal framework dedicated to artificial intelligence. It establishes harmonised rules across the EU to ensure AI systems are safe, transparent and respect fundamental rights.

The 4 risk levels

The AI Act classifies AI systems by risk level and imposes proportional obligations.

Unacceptable risk (prohibited): social scoring, real-time biometric identification in public spaces, etc.

High risk: biometric identification, recruitment, critical infrastructure, justice — strict obligations

Limited risk: chatbots, deepfakes — transparency obligations

Minimal risk: spam filters, AI video games — no specific obligation

Your AI Act obligations

Obligations depend on your role (provider, deployer, importer, distributor) and the risk level of the AI system.

Risk management system throughout the AI lifecycle

Technical documentation of the AI system

Records & logs of automated operations

Transparency towards users

Human oversight at all stages

Accuracy, robustness and cybersecurity of the system

Data governance for training, validation and testing

Sanctions

Sanctions are very heavy and based on global annual revenue.

Prohibited AI: up to EUR 35M or 7% of global revenue

Non-compliant high-risk AI: up to EUR 15M or 3% of global revenue

Inaccurate information: up to EUR 7.5M or 1% of global revenue

Simplifiez votre conformité AI Act avec Data Comply One

FAQ AI Act

The AI Act applies progressively between 2025 and 2027, with bans (unacceptable AI) effective from February 2025 and full high-risk obligations from August 2026.