Where Bulgarian business really stands on AI: 8.55% against the EU's 19.95%

According to Eurostat, in 2025 8.55% of Bulgarian enterprises with 10 or more persons employed used at least one artificial-intelligence technology, against an EU-27 average of 19.95% — less than half the Union average, and the third-lowest share among the 27 Member States, with only Romania (5.21%) and Poland (8.36%) ranking lower. More telling than the gap itself is its direction: in 2021 it stood at 4.36 percentage points; in 2025 it is 11.40. It is not closing — it is widening.

What the figure actually measures

8.55% does not mean "all Bulgarian companies". The Eurostat indicator (dataset isoc_eb_ai) covers enterprises with 10 or more persons employed — employees and self-employed — in NACE Rev. 2 aggregate C10-S951_X_K, that is "all activities except agriculture, forestry and fishing, and mining and quarrying", and excluding the financial sector. Micro-enterprises with fewer than 10 persons employed are outside the statistic altogether; no Bulgarian data are published for them.

"Using AI" means using at least one of eight technologies: text mining; speech recognition; natural language generation and speech synthesis, including programming code; generation of pictures, video or sound/audio; image recognition; machine learning, including deep learning, for data analysis; AI-based software robotic process automation; and technologies enabling machines to move physically by taking autonomous decisions. The data were collected by national statistical authorities in the first months of 2025.

The trend, with the necessary caveat

Bulgaria's series runs 3.29% (2021), 3.62% (2023), 6.47% (2024) and 8.55% (2025). The EU-27 equivalents are 7.65%, 8.06%, 13.48% and 19.95%. There is no 2022 observation, and none should be interpolated. One methodological caveat matters: until 2024 the indicator aggregated seven technologies; from 2025 it aggregates eight, after image, video and audio generation was added. Part of the 2024-to-2025 jump reflects the widened definition rather than adoption alone. The ratio is still instructive: over one year the EU added 6.47 points, Bulgaria 2.08. Measured against the EU average, Bulgaria's relative position has barely moved — 43.0% of the EU level in 2021 and 42.9% in 2025, with a peak of 48.0% in 2024 — while in absolute terms the distance it has to close has more than doubled.

Not merely a question of company size

The convenient explanation is that the Bulgarian economy is made of small firms. The data refute it. Across the EU, adoption rises from 17.0% among enterprises with 10-49 persons employed, through 30.36% for 50-249, to 55.03% for 250 and more. In Bulgaria the same bands are 7.17%, 13.3% and 26.18%. Even Bulgaria's largest enterprises therefore sit below the European rate for medium-sized firms, and at under half the European large-firm rate. The gap widens with company size — which points to capability, skills and management choice, not only to economic structure.

Slovenia shows the region is not destiny, at 21.61%, above the EU average. Slovakia is at 18.00%, Czechia 17.60%, Croatia 15.19%, Hungary 10.37% and Greece 8.93%. The leaders are Denmark at 42.03%, Finland 37.82% and Sweden 35.04%.

What holds enterprises back

Here the denominator decides everything. In 2025 only 11.1% of EU enterprises with 10 or more persons employed, and 4.89% of Bulgarian ones, had ever even considered using an AI technology. Among those that considered it, the EU reasons for not using AI were: lack of relevant expertise, 70.31%; lack of clarity about the legal consequences, 53.61%; concerns about breaching data protection and privacy, 52.72%; AI not useful for the enterprise, 17.79%. Bulgaria's figures are 72.69%, 50.29%, 47.68% and 16.26%. These shares are of enterprises that ever considered AI, not of all enterprises — quoted without that qualifier they mislead.

The 2024 comparison at EU level is revealing: among enterprises that had ever considered AI, "not useful" falls from 20.68% to 17.79%, while concerns about breaching data protection and privacy rise from 48.83% to 52.72%. Within that group, doubts about usefulness are receding while hesitation over the legal consequences and over personal-data protection is growing.

Employees are ahead of employers

A separate Eurostat survey — on the use of artificial intelligence by individuals — records that 22.50% of individuals aged 16-74 in Bulgaria had used generative AI tools in the three months before the survey, against 32.66% across the EU-27. In the same dataset (isoc_ai_iaiu, broken down by age group — see the note on sources), the share among Bulgarians aged 16-24 is 49.98%, against 26.78% for 25-54 and 8.09% for 55-74. The two surveys have different populations and their values cannot be subtracted from one another. The direction, however, is unambiguous: people are taking up these tools faster than organisations are formally adopting them.

That has a direct legal dimension. Article 4 of Regulation (EU) 2024/1689 (the AI Act) requires providers and deployers of AI systems to "take measures to ensure, to their best extent, a sufficient level of AI literacy of their staff and other persons dealing with the operation and use of AI systems on their behalf". The obligation has applied since 2 February 2025. In its non-binding guidance, the European Commission takes the view that a company whose employees use ChatGPT for, say, advertising copy or translation falls within scope: "Yes, they should be informed about the specific risks, for example hallucination." Supervision and enforcement rest with national market surveillance authorities and begin on 2 August 2026.

A note on sources

Every figure comes from Eurostat datasets isoc_eb_ai and isoc_ai_iaiu (updated 15 and 5 June 2026), extracted on 22 July 2026, and from the text of Regulation (EU) 2024/1689. The percentage-point differences and the ratios to the EU average are calculated directly from those published values. The age breakdowns for Bulgaria and the EU-27 come from isoc_ai_iaiu, indicator I_IUAI, categories Y16_24, Y25_54 and Y55_74; the exact extraction address is in the source list. Article 4 is quoted from the official English text of the Regulation on EUR-Lex, and the Commission's answer is published in English only. The enterprise data exclude micro-firms and the financial sector; the individual data cover ages 16-74. This text is not legal advice.