More than six in ten Romanian companies with at least 10 employees say they use artificial intelligence (AI). Yet only two in ten have done anything concrete to prepare their people for it. That is the finding of the study "Widespread use, shallow integration. Artificial intelligence in Romanian companies", launched by Edge Institute and the Concordia Employers' Confederation, one of Romania's main business associations.

The figures land on an already uncomfortable backdrop. According to Eurostat, the EU's statistics office, Romania ranks last in the European Union for AI use in companies, at 5.2%. The survey behind the new study was carried out by MIA between 30 June and 28 July 2026. It consisted of 1,003 telephone interviews with decision-makers at companies with at least 10 employees. It shows that the problem is not a lack of interest. The problem is how the technology does or does not reach the people doing the work.

For you as an employee or candidate, the consequence is direct. At many companies, if you want to learn to work with AI, you do it on your own. For HR, it is a sign that the statement "we use AI" says very little about what actually happens inside teams.

What "we use AI" really means

The 61.7% of companies that use AI in some form shrink quickly once you look at how often and how deeply they use it. Only 13.5% use it regularly, and 3.3% have built it into company processes. The rest could mean almost anything, from one employee drafting emails with a chatbot to a few experiments run out of curiosity.

The more important number for the people who work there is this one: in 56% of companies that use AI, the technology reaches fewer than a quarter of employees. Even at companies that tick the "AI" box, most staff don't actually work with it.

Only 2 in 10 companies train their people

For employees, training matters most. Only 20% of companies have any measure to prepare staff for AI. 26% have some kind of responsible-use rule, and 15% consider AI an important part of their strategy.

Put these figures side by side and you get a two-speed labour market. At a minority of companies, AI comes with training, clear rules and a place in the strategy. At most others, the technology is used informally by a handful of people with no guidance. Anyone who wants to keep up teaches themselves, often in their own time, and without knowing exactly what they are allowed to enter into an AI tool.

The absence of rules is not a technical detail. When nearly three-quarters of companies have no responsible-use policy, the risk of a mistake, such as customer data pasted into a public tool, falls on the employee rather than the organisation.

The underlying problem: digitalisation, not AI

The study shows that AI does not grow in a vacuum. 46% of companies use none of the eight digital tools measured in the survey. The gap by company size is huge: 55% of companies with 10 to 25 employees use none of them, compared with just 3% of large companies.

The link with AI is direct. AI use rises from 33% at companies with no digital tools to 91% at those using three or more. Alexandru Dănescu, co-author of the study, sums it up:

"The Romanian economy does not suffer from a lack of interest in artificial intelligence (AI), but it does have a digitalisation problem. Almost half of companies are not digitalised at all or only poorly, so in these organisations AI means at most a chatbot used by one employee. That is precisely why public funding in this area must start with basic digitalisation before any effort to integrate AI into companies' activities."

For an employee at a small company, this means the gap with peers at large companies is not only about AI. It is about the whole digital environment in which they build their skills, day after day.

For employees and candidates: how to position yourself

You can't change your company's strategy, but you can decide how you respond to it. A few things you can start doing tomorrow:

• Ask better questions at interview: "Do you use AI?" tells you little. It is more useful to find out how many people on the team actually work with it, whether there is training, and whether there are written rules on what data you may use.

• Check from the inside: the experiences of former and current employees often reveal more than a job ad about how much a company invests in training.

• Learn methodically, not at random: if your employer doesn't train you, pick a task tied directly to your work (reports, data analysis, writing) and practise on it, without using confidential information.

• Ask for rules where there are none: ask your manager which tools are allowed. A clear answer protects you, not just the company.

• Show what you know on your CV, concretely: instead of "AI skills", describe which task you sped up or which process you simplified with it.

If you already work at a company that gets this right (or, on the contrary, leaves you alone with the tools), your experience helps people now looking for jobs know what to expect.

For HR and employers: from informal use to team capability

For HR, the study's figures describe a risk rather than an achievement. A company where AI reaches fewer than a quarter of employees, with no training and no rules, has adoption that depends on a few individuals. When they leave, the know-how leaves with them.

A few directions worth assessing:

• Rules before tools: a short policy on which tools are permitted and which data must never leave the company. Only 26% of companies have one, so this is an easy gap to close.

• Training for the whole team, not just enthusiasts: with only 20% of companies offering any training, an employer that does it systematically also stands out to candidates.

• Basic digitalisation first: the study recommends starting with a digital tool for core functions, such as electronic document management.

• Measure real access: how many people use AI in their daily work says more than the fact that the company "has" AI.

On the money side, the intention is there; the funding less so. 30% of companies want to invest in AI, and 78% of them need external financing. Interest tracks digitalisation: only 6% of companies with no digital tools want to invest, compared with 75% of those using five or more. At the other end, 39% of companies have not looked into AI and do not consider it relevant.

The government programme for 2026 to 2028 announces vouchers for SMEs covering AI, cloud and cybersecurity, but with no budget and no timeline. Bogdan Dumitru, Romania's delegate minister for digitalisation and AI, explained the thinking behind the measure:

"As set out in the government programme, these vouchers are first and foremost meant to help a company, especially SMEs, validate a process, validate a particular integration, an adoption of AI components in their ecosystem. If this process works, it will continue to receive this kind of support."

Meanwhile, the payment deadline under the "Digitalisation of SMEs" scheme in Romania's PNRR (the National Recovery and Resilience Plan, funded by the EU) was 30 October 2026. For companies counting on public money in the coming period, there are, for now, more questions than answers.

Key takeaways

The study's conclusion is not that Romanian companies ignore AI, but that they use it on the surface without bringing their people along. For employees, AI skills currently come more from personal initiative than from internal programmes. For employers, the difference between "having" AI and truly using it starts with unglamorous basics: core digital tools, clear rules and training hours for the whole team.