Eurostat & European Commission

Who's actually using AI at work in Europe?

More than half of Europeans have tried AI, but only 27% use it for work, and those who do are convinced: 91% say they finish tasks faster. So which countries, company sizes and job functions are pulling ahead, and who is being left behind? Drawn from Eurostat's ICT usage survey and the European Commission's Spring 2026 Economic Forecast.

Reference years 2025 (enterprise & individual use) and 2026 (occupation survey). Compiled 29 September 2026.

EU enterprises using AI, 2025

20.0%

Up 6.5 pp vs. 2024

EU individuals using AI (any use)

54%

Use AI specifically for work

27%

Report faster task completion

91%

Avg. hours saved per month at work

7.4hrs

AI use has more than doubled since 2021, almost all of it since 2023

Share of EU enterprises (10+ employees) using AI technologies. The dataset has no 2022 figure.

Eurostat isoc_eb_ai

0%
10%
20%
Share of EU enterprises using AI, by year
YearShare
20217.7%
20238.1%
202413.5%
202520.0%

Big companies are over 3x as likely to use AI as small ones

Share of enterprises using AI, by number of employees, 2025

Small, 10–49 staff17.0%
Medium, 50–249 staff30.4%
Large, 250+ staff55.0%
All companies20.0%

Sales and marketing come first; logistics barely registers

Among companies that use AI, the share applying it to each business function, all sizes, 2025

Eurostat isoc_eb_ai

Marketing or sales34.7%
Business administration or management31.0%
Accounting, controlling or finance23.1%
Production processes20.8%
ICT security19.8%
R&D or innovation19.1%
Logistics6.1%

6 in 10 tech and media firms use AI. In construction, 1 in 10

Share of enterprises using AI, by economic sector, 2025. Tech and media is Eurostat's information & communication sector.

Eurostat isoc_eb_ain2

Information & communication62.5%
Professional, scientific & technical40.4%
Real estate24.8%
Administrative & support services19.9%
Wholesale & retail trade18.6%
Manufacturing17.3%
Accommodation & food service12.0%
Construction10.8%

Denmark 42%, Romania 5%: an 8x gap inside the EU

Share of enterprises (10+ employees) using AI, by member state, 2025. Each dot is a country: hover over or tap one to see its rank. Every value is listed below.

Eurostat isoc_eb_ai

EU average 20.0%
Romania 5%Denmark 42%
  1. 1Denmark42.0%
  2. 2Finland37.8%
  3. 3Sweden35.0%
  4. 4Belgium34.5%
  5. 5Luxembourg33.6%
  6. 6Netherlands33.2%
  7. 7Austria30.0%
  8. 8Germany26.0%
  9. 9Estonia23.4%
  10. 10Slovenia21.6%
  11. 11Malta21.5%
  12. 12Lithuania21.3%
  13. 13Spain20.3%
  14. 14Ireland19.6%
  15. 15France18.2%
  16. 16Slovakia18.0%
  17. 17Czechia17.6%
  18. 18Italy16.4%
  19. 19Croatia15.2%
  20. 20Latvia12.2%
  21. 21Portugal11.5%
  22. 22Hungary10.4%
  23. 23Cyprus9.3%
  24. 24Greece8.9%
  25. 25Bulgaria8.6%
  26. 26Poland8.4%
  27. 27Romania5.2%

The hours saved are similar. The job worries are not.

Monthly hours saved from AI use, by occupation group. EC Spring 2026 Forecast, n=21,207 across 18 member states and 4 candidate countries.

Managers & professionals

8.5hrs / month

Lower displacement concern

Highest work-related AI adoption of any group, and the largest share reporting a major productivity gain.

Plant & machine operators, elementary occupations

8.3hrs / month

Highest displacement concern

Time savings nearly match managerial roles, but this group is the most worried about AI replacing their job.

Students & unpaid workers

9.0hrs / month

High displacement concern

Largest time savings of any group, used mainly for study and job hunting rather than paid work. 41% report a quality improvement, 18% report none.

Generative AI is still mostly personal: 78% use it privately, 47% for work

Share of EU generative AI users, by purpose, EU-27, 2025. People can use it for more than one purpose.

Eurostat isoc_ai_iaiu

Private purposes

78.0%

Professional or work purposes

47.0%

Formal education

28.5%

Using AI safely at work

The adoption numbers are half the story. The other half is using these tools without leaking anything you shouldn't. That's what the 12-slide guide covers:

  • Why free, paid and business accounts treat your data differently
  • What counts as sensitive, with a before-and-after example
  • Why AI makes things up, and one prompt that helps