Tools Hub/AI Act Classifier

EU AI Act Risk Classifier

Pick your AI use cases for a directional Prohibited / High / Limited / Minimal tier and the obligations it triggers — a starting point for internal planning and EU customer conversations.

Step 1 · Pick the use cases (multi-select)

ProhibitedProhibited use case: cannot be placed on the EU market. The original Art. 5 prohibitions apply from 2 Feb 2025; the two prohibitions added by the AI Omnibus (non-consensual intimate content, CSAM) apply from 2 Dec 2026.

High-RiskHigh-risk: full conformity assessment and registration before placing on the market. After the AI Omnibus, Annex III high-risk rules apply from 2 Dec 2027 and Annex I embedded high-risk from 2 Aug 2028 — the deadline moved, the obligations did not, and readiness is usually a multi-quarter programme.

Limited RiskLimited risk: the Art. 50 transparency duties, applicable from 2 Aug 2026; systems generating synthetic content that were already on the market must be retrofitted for marking by 2 Dec 2026. Note that transparency duties carry no log-retention period — that is a high-risk obligation.

Minimal RiskMinimal risk: no mandatory duties; voluntary alignment recommended.

Step 2 · Model / system modifiers

General-Purpose AI (GPAI)Foundation model / multi-purpose LLM — triggers GPAI transparency dutiesTraining compute ≥ 10²⁵ FLOPsAbove the AI Act systemic-risk threshold — extra evaluation & notificationProcesses biometric / health dataAuto-escalates to High-Risk

Classification Result

Minimal Risk

Minimal risk: no mandatory duties; voluntary alignment recommended.

Prohibited

High-Risk

Limited Risk

Minimal Risk

This classification is derived from the use cases and model attributes you selected. It is a directional self-assessment — not legal advice, and not a substitute for conformity assessment or registration. The same model may fall into different tiers depending on its actual use and your role (provider or deployer); the final determination must follow the product’s real purpose and official guidance, reviewed by qualified professionals.

Obligations for this tier

Join EU AI voluntary code of conduct
Internal model cards + lightweight transparency log
Track GPAI systemic-risk threshold updates

Next steps

Request access to assessments Open the Transfer Matrix Contact an advisor

The application timeline this classification is based on

See all 28 obligations and sources

2024-08-01
Regulation enters into force
2025-02-02
General provisions and prohibited practices apply
2025-08-02
GPAI model obligations and governance framework apply
2026-08-02
General date of application; Article 50 transparency rules apply; enforcement starts
2026-12-02
Two new prohibitions apply; Article 50(2) retrofit deadline for legacy synthetic-content systems
2027-08-02
Legacy GPAI models must be compliant; national regulatory sandboxes operational
2027-12-02
Rules for Annex III high-risk AI systems apply (as postponed by the AI Omnibus)
2028-08-02
Rules for Annex I embedded high-risk AI systems apply (as postponed by the AI Omnibus)
2030-08-02
Compliance deadline for legacy high-risk systems intended for use by public authorities

Note: the GPAI obligations in Chapter V are a separate track from the four risk tiers above — one product can be a limited-risk system and a GPAI provider at the same time.

Filter the full obligation list for our business

Content as of 2026-09-12 · corpus 2026.09|Compiled from public official sources with links and review dates for compliance decision support. Rules change; have licensed professionals review before external disclosure, regulatory filing or contractual commitments.