The EU AI Act: get an overview of AI use in your company

An AI assistant for drafting copy and a system that helps screen job applicants are two different cases. Both can save work. A mistake in screening could cost an applicant a job.

The EU AI Act sets requirements according to the risks involved in using AI. The first question is simple: what does the system do, and who could be affected by an error?

An overview that supports decisions

A shared overview can bring together information otherwise held by individual employees or departments. It need not be extensive, but it needs to describe actual use.

• Task: What is AI used for, and what improvement do you expect?
• Data: What information can the system access?
• Consequences: Who is affected if an answer or action is wrong?
• Responsibility: Who follows up on its use and deals with errors?
• Checks: Who reviews the result before it is used?

A hypothetical example: marketing uses an AI assistant to draft copy from public product information. An editor checks the text before publication. Both belong in the overview so it shows how the work happens.

If the same tool is later used to assess job applicants, that use needs a separate check. The tool is the same. The job has changed.

What do the rules require of you?

Among other things, the Act requires companies that provide or use AI systems to take measures to support their staff’s AI literacy. In practice, this means helping people who use AI at work gain enough knowledge to use it sensibly. The overview is a useful place to begin because it shows which tasks they need to handle.

The Act also includes transparency requirements for certain systems and certain types of AI-generated content. The transparency requirements in Article 50 apply from 2 August 2026.

The specific requirements depend on what you use AI for and your company’s role. That role includes whether you use an AI solution or develop and sell one yourselves. For most companies like yours, using a solution will be the starting point.

The overview helps identify questions that need clarification. It is not proof of compliance.

If a system influences decisions about people, start there. Its use also deserves a closer look if you are unsure whether customers or employees need to be told about it.

Give someone responsibility for the overview

Naming a responsible person for each use makes follow-up easier. That person should be able to explain its purpose, locate the relevant agreements and ensure unresolved questions are addressed.

Employees need knowledge suited to their tasks. Someone using AI to draft copy, for example, needs to recognise fabricated information and know what they may share with the tool.

Revisit the overview when a department changes how it uses AI or gives a system access to new data. Otherwise, it can soon describe a company that no longer exists.

It may also reveal that several departments are paying for tools that serve the same purpose. Management can then decide whether the expense is justified and whether a shared solution would make work easier.

If several employees or departments use AI, a shared overview gives management something concrete to work with: responsibility for each use and a starting point for finding unnecessary spending or taking a closer look at how AI is used.