Clearing up Omnibus deadlines

Q: Deadlines for AI systems you already use (as of August)

Live: AI literacy requirements, for both deployers and providers2 December 2026: Transparency requirements for certain systems - those interacting with people, generating text or images, producing text used to inform the public, or generating deepfakes2 August 2027: Requirements for providers of general-purpose AI models that were on the market before August 20252 December 2027: Requirements for providers of high-risk systems in sensitive areas, e.g. hiring and healthcare2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices2 August 2030: Requirements for deployers and providers of high-risk systems intended to be used by public sector authoritiesNote: High-risk systems already on the market do not need to retroactively comply with high-risk requirements, unless their design has changed significantly. This exception probably won't apply to a lot of systems - meaning there won't be a lot of unchanged-since-2026 AI systems - but it is there. From the Act: This Regulation shall apply to operators of high-risk AI systems... that have been placed on the market or put into service before 2 August 2026, only if, as from that date, those systems are subject to significant changes in their designs.(Article 111(2))

Q: Deadlines for an AI system put on the market in August or later

Live: AI literacy requirements Live as of 2 August: Transparency requirements2 August 2027: Requirements for providers of general-purpose AI models that were on the market in August 20252 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices

Q: Deadlines for high-risk AI systems

2 December 2027: Requirements for providers of high-risk systems used in sensitive areas, e.g. hiring and healthcare2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices31 December 2030: Requirements for high-risk systems intended to be used by public authorities.

Q: Deadlines to meet transparency requirements

2 August 2026: for systems put on the market as of this date 2 December 2026: for systems already on the market, before 2 August 2026

Q: Deadlines for general purpose AI models

2 August 2025: For providers to meet all obligations, if their model was on the market after 2 August 20252 August 2025: For providers (or authorized representatives) to report if their model has systemic risk2 August 2027: For providers to meet all obligations, if their model was on the market before 2 August 2025

Q: Deadlines if you are a deployer

Live: AI literacy requirements2 August 2026: Transparency requirements if your system is an emotion recognition system or a biometric categorization system (not if it is used in the workplace - that would be prohibited), or if your system generates/manipulates text to inform the public, or if your system makes deepfakes

Q: Deadlines if you are a provider

2 December 2026: Transparency requirements if your system interacts with people or generates text/images/video2 August 2027: Requirements for providers of general-purpose AI models that were on the market before August 20252 December 2027: Requirements for providers of high-risk systems in sensitive areas, e.g. hiring and healthcare2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices

Q: Deadlines if you are outside the EU

2 August 2025: For providers of general purpose AI models to mandate an authorized representative 2 August 2025: For authorized representatives of general purpose AI models to report if their model has systemic risk 2 December 2027: For providers of high-risk systems in sensitive areas, e.g. hiring and healthcare, to mandate an authorized representative, and for authorized representatives to meet requirements2 August 2028: For providers of high-risk systems embedded in regulated products, e.g. medical devices, to mandate an authorized representative, and for authorized representatives to meet requirements

Q: Deadlines for AI systems used within "large-scale IT systems in the area of freedom, security, or justice"

2 August 2030: An evaluation of whether these systems meet AI Act requirements, if systems were on the market before 2 August 2027

Ley Muller
May 29
4 min read

Updated: Jun 1

6 deadlines for the EU AI Act post-Omnibus, two that have already passed, and those that are live already for Norwegian providers

An old calendar of the month of august, with a woman walking in the background — *Image: Cambridge Historical Society, CC BY*

Organized in multiple ways - expand the heading most relevant to use. These are not mutually exclusive categories; an AI system can have both high-risk requirements and transparency requirements, for example.

(I stopped writing "Live: AI literacy requirements", because these are valid for every category: high-risk, transparency risk, provider, deployer, etc.)

The first upcoming deadline is 2 August of this year. Would you like a quick check of what your deadlines are and how to meet them? We offer checks to both AI developers and to organizations buying/using AI.

Deadlines for AI systems you already use (as of August)

Live: AI literacy requirements, for both deployers and providers
2 December 2026: Transparency requirements for certain systems - those interacting with people, generating text or images, producing text used to inform the public, or generating deepfakes
2 August 2027: Requirements for providers of general-purpose AI models that were on the market before August 2025
2 December 2027: Requirements for providers of high-risk systems in sensitive areas, e.g. hiring and healthcare
2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices
2 August 2030: Requirements for deployers and providers of high-risk systems intended to be used by public sector authorities

Note: High-risk systems already on the market do not need to retroactively comply with high-risk requirements, unless their design has changed significantly. This exception probably won't apply to a lot of systems - meaning there won't be a lot of unchanged-since-2026 AI systems - but it is there. From the Act:

This Regulation shall apply to operators of high-risk AI systems... that have been placed on the market or put into service before 2 August 2026, only if, as from that date, those systems are subject to significant changes in their designs.

(Article 111(2))

Deadlines for an AI system put on the market in August or later

Live: AI literacy requirements
Live as of 2 August: Transparency requirements
2 August 2027: Requirements for providers of general-purpose AI models that were on the market in August 2025
2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices

Deadlines for high-risk AI systems

2 December 2027: Requirements for providers of high-risk systems used in sensitive areas, e.g. hiring and healthcare
2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices
31 December 2030: Requirements for high-risk systems intended to be used by public authorities.

Deadlines to meet transparency requirements

2 August 2026: for systems put on the market as of this date
2 December 2026: for systems already on the market, before 2 August 2026

Deadlines for general purpose AI models

2 August 2025: For providers to meet all obligations, if their model was on the market after 2 August 2025
2 August 2025: For providers (or authorized representatives) to report if their model has systemic risk
2 August 2027: For providers to meet all obligations, if their model was on the market before 2 August 2025

Deadlines if you are a deployer

Live: AI literacy requirements
2 August 2026: Transparency requirements if your system is an emotion recognition system or a biometric categorization system (not if it is used in the workplace - that would be prohibited), or if your system generates/manipulates text to inform the public, or if your system makes deepfakes

Deadlines if you are a provider

2 December 2026: Transparency requirements if your system interacts with people or generates text/images/video
2 August 2027: Requirements for providers of general-purpose AI models that were on the market before August 2025
2 December 2027: Requirements for providers of high-risk systems in sensitive areas, e.g. hiring and healthcare
2 August 2028: Requirements for providers of high-risk systems embedded in regulated products, e.g. medical devices

Deadlines if you are outside the EU

2 August 2025: For providers of general purpose AI models to mandate an authorized representative
2 August 2025: For authorized representatives of general purpose AI models to report if their model has systemic risk
2 December 2027: For providers of high-risk systems in sensitive areas, e.g. hiring and healthcare, to mandate an authorized representative, and for authorized representatives to meet requirements
2 August 2028: For providers of high-risk systems embedded in regulated products, e.g. medical devices, to mandate an authorized representative, and for authorized representatives to meet requirements

Deadlines for AI systems used within "large-scale IT systems in the area of freedom, security, or justice"

2 August 2030: An evaluation of whether these systems meet AI Act requirements, if systems were on the market before 2 August 2027

One AI system can fall into multiple categories, each with separate deadlines.

An independent check of your deadlines, according to whether you are a provider or deployer, the intended use of the AI system, the risk class, and when you started using it, is a quick way to make sure you are on track with compliance.

Book a meeting, or request more info