Skip to content

EU regulation

EU AI Act

The European regulation for artificial intelligence. Obligations differ per risk category; below are the categories with what they mean and when they apply, each traceable to the legal text.

DepthWorked out, with practical questions on HR

The AI Act sets duties per AI system, depending on its risk and your role. Trusq records your AI systems and works out the duties that apply, with the legal text per article and practical questions on AI in HR (in Dutch). Trusq does not decide whether a system complies.

Does this apply to you?

  • You develop, place on the market, import, distribute or use an AI system in the EU; your role determines your duties: provider, deployer, importer or distributor (Art. 2(1) and Art. 3).
  • The risk of the use determines how heavy the duties are: prohibited practices (Art. 5), high-risk AI (Annex III and Annex I), transparency obligations (Art. 50) and AI literacy (Art. 4).
  • The duties of deployers do not apply to natural persons who use AI systems in the course of a purely personal non-professional activity (Art. 2(10)).

Risk categories and obligations

Prohibited practice (Art. 5)
This use is outright banned in the EU — it cannot be mitigated, only stopped. The highest fines in the regulation.
High risk (Annex III/I)
The heaviest set of duties: risk management, data governance, documentation, human oversight, conformity assessment.
GPAI model (Chapter V)
A dedicated regime for providers of general-purpose AI models; extra duties for systemic risk (training above 10^25 FLOP).
GPAI in use
You build on a third-party AI model (such as GPT, Claude or Gemini). The GPAI regime targets that model's provider — but you must be able to rely on their compliance (vendor assessment) and, where outputs reach the public, follow the transparency duty (Art. 50). Note: if you substantially modify the model and place it on the market under your own name, you may become a provider yourself.
Transparency (Art. 50)
A duty to make AI recognisable: disclose chatbots, mark synthetic output machine-readably, label deepfakes.
AI literacy (Art. 4)
Anyone developing or deploying AI must take measures that support AI literacy among staff who operate it.
Minimal risk
Most AI applications fall here; voluntary codes of conduct are encouraged.

Deadlines

Prohibited practice (Art. 5)in force since 2 February 2025; two new prohibitions apply from 2 December 2026 (Regulation (EU) 2026/1744)
High risk (Annex III/I)Annex III: 2 December 2027 (was 2 August 2026); Annex I products: 2 August 2028 (was 2 August 2027)
GPAI model (Chapter V)in force since 2 August 2025; AI Office enforces from 2 August 2026
GPAI in usethe heavy GPAI duties sit with the model provider; for you, vendor assurance and transparency apply
Transparency (Art. 50)applies from 2 August 2026; only Article 50(2) marking for systems placed on the market before that date applies from 2 December 2026
AI literacy (Art. 4)in force since 2 February 2025
Minimal riskno specific AI Act obligations beyond AI literacy

Full detail, sources and changelog: AI Act deadlines — verified against the source →

Official text, structured by article

Browse all 113 articles, 180 recitals and 13 annexes from the official EUR-Lex text. Each page carries its source version and keeps proposed changes separate from current law. Provisions inserted by Regulation (EU) 2026/1744 — among them Articles 4a, 60a and 75a and Annex XIV — have no page of their own; the amending act lists them.

Open the AI Act Article Explorer →

What Trusq supports, and what not

Trusq supports

  • Record AI systems with purpose, role and owner, and see the classification and the duties that follow from it, with their source.
  • The transparency obligations of Article 50 and the duties of the deployer as actions with expected evidence.
  • Changes to the AI Act and its guidelines, traced to your systems.

Not supported

  • No conformity assessment and no registration in the EU database on your behalf.
  • Trusq does not decide whether your system complies; you assess the outcome, with an adviser where needed.

Follow this topic

Get an email whenever something changes here. No account needed; confirm by email (double opt-in).

We use your email address only to send updates about this topic. You can unsubscribe at any time. See our privacy policy.

← Back to Trusq

Independent expertise

Need a specialist for this work?

Describe the framework, location and assignment parameters. Trusq reviews every request and makes the selection by hand; profiles are not public and no automated ranking is used.

Request an expert

Are you an expert?