The EU AI Act timeline
12th July 2024
The AI Act has been published in the
Official Journal of the European Union, meaning it will officially become
law after years of negotiation.
AI Act published
1st August 2024
The AI Act now applies but is not yet enforced, it will be phased out over
the next two years.
AI Act becomes law
2nd February 2025
Systems that are deemed unacceptable risk, generally based on their potential impact on civil liberties and society,
are banned and fines of 7% of global
turnover or €35 million will apply.
Unacceptable risk systems banned
2nd August 2025
The AI Office must work with the board to adopt codes of practice that will help organisations demonstrate compliance with the requirements.
Codes of practice from to be adopted
2nd February 2026
The Commission must issue guidance on
how High-risk AI systems are classified. While these are listed in the Annex's
the Commission has the power to
add more use cases as necessary.
High risk AI systems
2nd August 2027
Systems that were put on the market
prior to 2nd August 2025 must comply
with all the obligations relevant.
Compliance
Each country must notify the Office of the national competent authority, the market surveillance authority and the single point of contact.
Authorities appointed
Guidance on the reporting of serious incidents is to be published.
Serious incidents
Each country must notify the Office of the national competent authority, the market surveillance authority and the single point of contact. General Purpose AI systems need to meet the obligations listed in Chapter 5, such as submitting certain information about sues and training data to the AI Office.
General purpose AI requirements to be met
The commission must publish an implementing act to ensure that the
market surveillance of the products is uniform across the EU, and the national authorities have a template to follow.
Post market surveillance
Systems that are a safety
component of regulated products
in Annex I must comply
The Artificial Intelligence Act has been developed over several years with much commentary and debate. In early 2024, it reached a political agreement in the EU. The Act will be rolled out in a phased approach over the next two years, all obligations must be met by August 2026. Compliance with the Act will likely require some changes as organisations create their AI strategy and begin to roll out tools. Key elements include establishing risk and quality management systems, monitoring the outputs of the AI, ensuring the right level of AI literacy for the users of the systems, and maintaining the systems. Changes needed to meet the requirements will depend on existing governance frameworks and the uses of AI in the organisation but needless to say, there is a lot of work needed across the board to provide for the safe, responsible and effective use of AI.
Below, we detail the timeline that all organisations using AI need to be aware of:
Contact
Senior Manager, Consulting
David O' Sullivan
Contact
Related Pages
Risk consulting
Responsible Artificial Intelligence & EU AI Act Compliance
2nd August 2026
This is enforcement day, organisations must comply with the requirements relevant to them or face fines of 7%, 3% or 1% depending on the infringement.
Entry into force and application
Each country must establish a regulatory sandbox that will allow
for innovation and the development
of AI systems
AI regulatory sandbox to be in place
Firms are required to ensure AI Literacy of the staff and other persons dealing with the operation and use of AI systems.
AI Literacy