Enforcement begins August 2, 2026

EU AI Act Compliance Guide

The European Union's AI Act is the world's first comprehensive AI regulation. Here's what you need to know to ensure your AI agents are compliant.

Key dates

August 1, 2024

AI Act enters into force

February 2, 2025

Prohibited AI practices become enforceable

August 2, 2025

General-purpose AI rules apply

August 2, 2026

Full enforcement for high-risk AI systems

18 months away

Risk categories

The AI Act categorizes AI systems by risk level. Most business AI agents fall into the high-risk category.

Unacceptable Risk

Prohibited

Social scoring, real-time biometric identification in public spaces, manipulation of vulnerable groups. These are banned outright.

High Risk

Strict Requirements

AI systems that impact health, safety, or fundamental rights. Includes:

• Credit scoring and loan decisions

• Employment and HR decisions

• Educational assessments

• Law enforcement applications

• Migration and asylum

• Critical infrastructure

Limited Risk

Transparency Only

Chatbots, deepfakes, emotion recognition. Must disclose AI involvement to users.

Minimal Risk

No Requirements

Spam filters, AI-enabled video games. No specific obligations.

Article 14: Human Oversight

High-risk AI systems must enable human oversight. This is where Empress helps.

Article 14 requires:

1.Humans must be able to understand the AI system's capabilities and limitations

2.Humans must be able to correctly interpret the AI system's output

3.Humans must be able to decide not to use the AI system or override its output

4.Humans must be able to intervene or interrupt the AI system

How Empress helps:

Complete audit trail of every AI decision in xAPI format

Reasoning capture shows why decisions were made

Human override logging tracks when humans intervene

One-click compliance reports for auditors

Real-time dashboard for monitoring AI behavior

Penalties for non-compliance

€35M

or 7% of global revenue

Prohibited AI practices

€15M

or 3% of global revenue

High-risk violations

€7.5M

or 1% of global revenue

False information

Compliance checklist

Classify your AI systems by risk categoryEmpress helps

Implement human oversight mechanismsEmpress helps

Document AI decision-making processesEmpress helps

Enable audit trail and loggingEmpress helps

Conduct risk assessments

Appoint authorized representatives in EU

Prepare technical documentationEmpress helps

Implement quality management system

Enable data governance and traceabilityEmpress helps

Start now, be ready by August 2026

Get compliant with Empress

Complete audit trails, human oversight documentation, and one-click compliance reports. Join the beta to get started.

Join waitlist View compliance features →