AI compliance and training: the AI Act and the skills to develop

What does the regulation require?

Regulation (EU) 2024/1689, known as the AI Act, covers providers and deployers of AI systems. A deployer is any organisation that uses an AI system under its own authority, which includes an SME whose employees use a chatbot. No size threshold applies.

Regulation (EU) 2026/1744, known as the Digital Omnibus, rewrote Article 4. Companies must now take measures to "support the development of AI literacy", without being required to guarantee a given level for each person. The European Commission states that no standard training, number of hours or certificate is required. The details of the text, timeline and penalties are set out in AI Act Article 4: mandatory AI training, what companies must do.

Enforcement has been in the hands of national market surveillance authorities since 2 August 2026. In France, their designation depends on a bill still before Parliament as of 10 October 2026.

Where to start?

The first step is a stocktake: which AI tools are used, by whom, for which tasks and on which data. According to Insee (the French national statistics institute), 18% of French companies with 10 or more employees used at least one AI technology in 2025, compared with 10% a year earlier. This figure only measures uses declared by the company: unmanaged individual uses often go unnoticed.

This inventory also helps position the organisation: scattered uses, controlled deployment on a few processes, or integration at scale. The method is presented in AI maturity in companies: how to assess where you stand.

Which skills to train, and for whom?

The definition of AI literacy (Article 3 of the AI Act) combines skills, knowledge and an understanding of opportunities and risks. In practice, needs differ by role:

Population	Main need
All employees who use AI	Understand how tools work and their limits, know which data not to enter, check outputs
Regular users	Integrate AI into job tasks, assess the quality of results
Managers	Organise work, assess performance and support teams in an AI context
People in charge of human oversight of high-risk systems	Competence, training and authority required by Article 26 of the AI Act

The European DigComp framework, cited by Regulation 2026/1744 as a possible reference, now integrates AI across its 21 digital competences. Priorities by profile are detailed in Which AI skills to train first by role and sector, and the case of managers in Training managers to manage in an AI-augmented workplace.

Knowing the limits of AI is part of the literacy expected: see What AI cannot do in training.

How to document and fund it?

The law imposes no format of proof. The Commission indicates that an internal record of trainings and guidance initiatives can be kept. Measuring what people can do after training is not required by Article 4, but it is what shows whether the measures have had an effect: see Measuring skill acquisition beyond completion rates.

Under French law, AI training falls within the employer's obligation to adapt employees to changing technologies (Article L6321-1 of the Labour Code) and can be included in the skills development plan. Other training required by law is listed in Mandatory training in companies: what are your legal obligations. Available funding schemes (CPF, the individual training account; OPCO, the sector skills operators; FNE; Pro-A) are presented in Funding training in companies: the complete guide.