AI Literacy in 2026: An EU AI Act Readiness Guide for Organizations

AI is no longer an issue for technology teams alone. Employees in human resources, learning and development, operations and management roles interact with different systems for tasks ranging from content generation and data interpretation to candidate assessment and personalized learning. As this interaction increases, the question “Does everyone know how to use AI?” becomes “How do we measure, develop and monitor the knowledge and behaviors our organization needs?”

One of the key topics in 2026 is the EU Artificial Intelligence Act (AI Act). Article 4 of the Act provides that providers of AI systems and organizations using these systems must take measures to support AI literacy among staff and other people dealing with the systems on their behalf. As of August 2, 2026, supervision and enforcement of Article 4 are on the agenda of national market surveillance authorities.[2] This article is not legal advice; it is a practical framework for organizations designing their learning and development plans to meet current needs.

Why is AI literacy more than a training topic in 2026?

The European Commission does not define a single exam, a single course duration or the same “sufficient level” for everyone under Article 4. Instead, an employee’s technical knowledge, experience and training should be considered, along with the risks of the system used, its purpose and the people affected by it.[1] This approach carries an important message for corporate development teams: instead of one-off training completed with a general AI presentation, organizations need a continuous development pathway designed according to role and risk.

The need is not limited to regulation. According to the OECD’s 2026 AI and skills study, the skills shortage is one of the major barriers to AI adoption, especially in small and medium-sized enterprises. While the share of workers who need advanced AI expertise remains below 1%, using, analyzing and interpreting data, as well as problem-solving, creativity and management skills, remain important.[4] For this reason, AI literacy should not be treated solely as training in “writing good prompts.”

A role-based development model for AI Act readiness

The first step is to make AI touchpoints within the organization visible. One team may use generative AI to draft content, while another may use a system that generates assessments about candidates, employees or students. Offering everyone the same training content can overlook the difference in risk and responsibility between these two roles.

1. Inventory AI use

Which team uses which system, with what data and for what purpose? Is the output only a draft, or is it a recommendation that affects a person’s access to education, employment or development opportunities? The answers determine the scope of training and where human oversight should be established.

2. Define skill levels

Employees at the beginner level need basic concepts, data privacy and verification habits. At the management level, responsibility for decisions, explainability, bias risk and a culture of safe use within the team come to the fore. More detailed objectives, such as model limitations, logging, integration and security, can be added for technical teams.

3. Put skills into practice and generate evidence

Watching a video alone does not show that a skill has translated into behavior. Use role-based scenarios, short assessments, feedback and repeat measurement. For example, before sharing an AI-supported development recommendation with an employee, a manager should be able to distinguish the data, the assumptions and the human decision. This kind of evidence offers a more meaningful picture of development than training completion rates.

Turning development into a measurable process with Qulak

Qulak focuses on managing learning and development processes end to end and connecting employee potential with productivity. This makes it possible to address a current need like AI literacy as an interconnected development pathway rather than a single content campaign.

1. Baseline assessment: Use skills assessments to make teams’ levels of understanding, risk awareness and practical application visible.
2. Personal development map: Assign learning objectives and content based on role, use case and skill gaps.
3. Practice and support: Support real work scenarios with training programs, consulting or mentoring.
4. Repeat measurement: Use advanced assessment and development tracking to monitor whether knowledge transfer is translating into behavior.

The value of this process lies not only in producing reports, but in showing the organization which team, role and use case require action to close a skill gap. Tools such as personalized learning experiences, digital badges or certification can also make development visible. However, the purpose of a badge should not be merely to mark completion, but to link it to a measured skill and demonstrated behavior.

An actionable plan for the first 90 days

In the first 30 days, inventory AI use and identify risk priorities. For each role, document what employees “need to know,” “need to be able to do” and “must not decide.” At this stage, also clarify whether your organization is a provider, a user or both.

In days 31–60, launch role-based learning pathways. Alongside a general awareness module, work through topics such as data and privacy, verification, bias, human oversight and incident reporting using real-world scenarios. For use cases that may be high-risk, do not rely solely on instructions for use; the European Commission also emphasizes the importance of training and guidance suited to the context.[1]

In days 61–90, conduct a short repeat assessment and gather feedback from managers. Which team regularly verifies outputs? In which use cases is human approval unclear? Which skill gaps could affect access to work or learning? Carry these answers into the next development cycle. This way, AI literacy becomes an organizational competency that is measured and updated, rather than training that is forgotten by year-end.