Obligations
EU AI Act risk classification.
Regulation (EU) 2024/1689 (the EU AI Act) sorts systems by what they do and the context they do it in. The tier determines nearly everything that follows. A wrong tier does not produce slightly wrong obligations. It produces the wrong set of obligations.
The tiers, and what changes at each
Four tiers, and what each one means for you.
The tiers depend on use and context, not on model capability.
Prohibited practices
A small set of uses the Act does not permit at all. The question is not whether your system is sophisticated. It is whether its use falls inside one of the described practices.
High-risk
This tier carries the substantial obligations, including the deployer obligations and, for certain deployers, the fundamental rights assessment. The Act’s own criteria and annexes decide it, not a general judgment of seriousness.
Transparency-obliged uses
Interactive systems and generated or manipulated content attract disclosure obligations whether or not the system is high-risk. A chatbot can sit outside the high-risk tier and still owe a transparency obligation.
Minimal risk
All other systems. Keep a dated record that you considered the classification, because the other results depend on it.
This page cannot answer the question for you. It sets out the questions that decide it. Whether an obligation attaches to a particular AI use case depends on facts specific to your organization. Your counsel decides whether your position is adequate. Nothing here is legal advice. Classification depends on the specific system, its intended purpose and the context of use. This page describes the structure, not your system’s tier.
How LegisGate classifies
Fixed rules classify, not a language model.
Fixed rules, applied by software, compare your answers with the Act’s criteria. No language model takes part.
- 1
From your answers
You give the intended purpose, the context of use, the decision the system informs and the sector. LegisGate does not infer them from a product description.
- 2
Higher tier where the Act is ambiguous
Where the Act genuinely allows two readings, the classification errs toward the higher tier. This is deliberate policy. Over-classifying costs extra work, and under-classifying leaves obligations unaddressed.
- 3
Recorded either way
The classification and its basis are part of the record, so a reviewer can disagree with the reasoning and not only with the outcome.
The tier determines the rest
Classify before you assess.
Every other EU AI Act question depends on the tier, including the fundamental rights impact assessment (FRIA), deployer obligations and transparency. The tier is the first thing LegisGate establishes for an AI use case, which is one tool used for one purpose. Add an AI use case to your inventory to see its tier.