Half of companies have lost track of employees' AI use

Half of Danish small and medium-sized companies have lost track of how employees use artificial intelligence, abbreviated AI, in their daily work. That's according to a survey from the research firm Voxmeter, conducted for the IT company itm8. The survey is based on responses from 401 IT managers at companies with between 50 and 250 employees.

The numbers speak for themselves. At half of the companies, employees use AI without shared guidelines for what is and isn't allowed. Two out of three companies have no overall plan at all for how AI should be used. And even among those that actually have a plan, there are still 40 percent where employees in practice do as they please. More than half of respondents, 51.6 percent, are worried that confidential company information is being shared with AI tools without anyone knowing.

Itm8 calls the phenomenon EGO-AI. It means that each individual employee finds and evaluates their own AI tools, and that the knowledge and good habits built up stay with the person instead of benefiting the whole company. Christian Binggeli-Winter, a partner at itm8, points out that the company thereby misses out on the gains that arise when good solutions are shared across departments.

Kasper Lynge Jacobsen, chief AI analyst at the Danish Chamber of Commerce (Dansk Erhverv), goes a step further and calls it a failure of management rather than an employee problem. It doesn't help to have a plan on paper if it isn't followed in practice, he says.

That picture matches the Danish Chamber of Commerce's own figures from its member companies. Today, 70 percent of companies use AI, up from 44 percent in 2023. But only 31 percent have a clear plan with defined roles and responsibilities for who decides what. According to the Danish Chamber of Commerce, the biggest obstacle is neither money nor legislation, but a lack of knowledge among both managers and employees.

New EU rules ahead

There's good reason to take the problem seriously right now. The EU's new rules for artificial intelligence, the so-called AI Act, take full effect from August 2, 2026. The most serious violations can cost fines of up to 7 percent of a company's total global revenue. What used to be something you had time to sort out is becoming something that's expensive to ignore.

Five concrete solutions

Itm8 proposes five concrete steps for companies that want to get things under control.

First, you should map what employees actually use today, simply by asking them. A plan that doesn't match reality just ends up in a drawer.

Next, you should divide data into three categories: green for data that may be freely shared with AI, yellow for data that requires caution, and red for data that is strictly forbidden to share.

Third, the company should approve a short list of AI tools and make them easy to use. The harder it is to access an approved tool, the greater the risk that employees find their own solutions that bypass the IT department.

Fourth, you should appoint a responsible person and introduce a rule that all content produced by AI is always read through by a human before it leaves the building.

Finally, good solutions should be shared internally in the company instead of staying with the individual employee. That's the very leap from what itm8 calls EGO-AI to what it calls ECO-AI, where security and reuse are built in from the start.