EU AI Act guides

Practical, article-referenced answers to the questions every company shipping or deploying AI is asking since the high-risk obligations became enforceable on 2 August 2026. No signup, no fluff — and when you want your own systems classified, the free assessment takes about 10 minutes.

August 2, 2026: what just became enforceable under the EU AI Act

High-risk obligations and the penalty regime of the EU AI Act became applicable on 2 August 2026. What is now enforceable, who is exposed, and what to do first.

Read the guide →

The EU AI Act compliance checklist

A practical, article-referenced checklist for EU AI Act compliance: classify every system, map obligations by role and risk tier, and close the gaps in order.

Read the guide →

Is your AI system high-risk? Annex III, explained

The eight Annex III categories that make an AI system high-risk, the Article 6(3) derogation that can take you out of scope, and how to document either answer.

Read the guide →

Annex IV technical documentation: what it is and what goes in it

Every high-risk AI provider must hold Annex IV technical documentation before placing a system on the EU market. The required sections, common gaps, and how to draft it fast.

Read the guide →

EU AI Act fines: the complete penalty guide

€35M or 7% for prohibited practices, €15M or 3% for breached obligations, €7.5M or 1% for misleading regulators — who fines whom, since when, and how exposure is calculated.

Read the guide →

You don't build AI — you use it. Deployer obligations under Article 26

Buying AI from a vendor does not transfer compliance. The duties Article 26 puts on companies that deploy high-risk AI: oversight, input data, monitoring, logs, worker notification.

Read the guide →

The FRIA: who must run a fundamental rights impact assessment

Public bodies, public-service providers, and deployers of credit-scoring and insurance-pricing AI must complete a FRIA before first use. What Article 27 requires, section by section.

Read the guide →

GPAI under the AI Act: Articles 53 and 55 for model providers

General-purpose AI model providers have had duties since August 2025: technical documentation, downstream information, copyright policy, training-data summaries — and more at systemic risk.

Read the guide →