The EU revised its AI Act, removing specific AI mastery requirements for staff, shifting oversight to national authorities and allowing companies to tailor training to their needs.
EU revises AI training requirements
The European Union has updated its AI Act, removing the requirement for companies to ensure their staff achieve a specific level of AI mastery. Previously, the law required that staff using AI systems have "as far as possible, a sufficient level of AI mastery," but this has now been replaced with a more flexible approach.New regulation focuses on promoting AI mastery
The revised regulation, published in July 2026, emphasizes promoting the development of AI mastery without defining a specific level for each employee. The European Commission clarified in an updated FAQ that both providers and users of AI systems are still obligated to provide training, but no formal certificate is required.Responsibility shifts to national authorities
National market surveillance authorities are now responsible for enforcing the new AI training rules. These authorities have been tasked with ensuring compliance since August 2, 2026, and can impose proportional sanctions if companies fail to meet the requirements.Some companies may misinterpret the change
Some companies, especially small and medium-sized enterprises and law firms, might mistakenly believe the new regulation allows them to skip training. However, the legal obligation remains, though companies can now tailor training to their specific needs and circumstances.AI training remains crucial for compliance
According to a report by Insee, 18% of companies with 10 or more employees used AI technology in 2025, up from 10% the previous year. Many of these companies reported being hindered by a lack of expertise and concerns about data protection, highlighting the continued importance of proper AI training.Training focuses on four key areas
The European Commission emphasizes that the risk of AI misuse remains the same, as it often stems from everyday practices rather than the tools themselves. The new regulation allows companies to tailor training to four main areas: understanding AI assistants, handling data, verifying information, and defining acceptable uses of AI tools.
