EU Small Firms Use AI to Monitor Staff but Distrust Its Calls

Forty-two per cent of European small, medium and mid-cap companies already use artificial intelligence or algorithmic management, and 36 per cent of the firms interviewed apply it to monitoring performance and productivity. Almost half, 49 per cent, say they do not trust the same technology for decisions about their staff. The European Commission published those findings on 14 October 2026.

The numbers come from a Flash Eurobarometer carried out between March and April 2026 across all 27 member states, based on around 12,826 telephone interviews with the heads of privately owned SMEs and mid-caps employing between one and 749 people. A further 500 interviews were conducted in the larger member states and 250 in the smaller ones. The Commission's own description of the adoption it measured is that it is "mostly in a limited and experimental way".

Where the algorithms actually sit

Algorithmic management, in the Commission's definition, covers digital systems that help managers organise a workforce or support decisions about tasks, schedules, performance or staffing, whether or not AI drives them. That bundling matters when reading the use figures: customer service and client relations lead at 47 per cent, performance and productivity monitoring follows at 36 per cent, and logistics, planning and staff scheduling at 30 per cent. Two of those three categories point at employees rather than customers.

Set against the adoption figure, two other results do most of the work. Ninety per cent of the companies surveyed said people take the decisions, and 2 per cent of decisions are fully automated. Sixty-one per cent said AI helps them take decisions, and 75 per cent expect AI and algorithmic management to improve job quality.

The honest reading is that the technology is in the workflow and not in charge. A firm that uses an algorithm to draft a rota and a manager who signs it off both appear in the 42 per cent, and the survey offers no measure of depth to separate them. The 2 per cent figure is the closest thing to a ceiling on how much decision authority has actually moved.

Three barriers inside two points of each other

Asked what holds them back, companies named high investment costs at 50 per cent, legal uncertainty at 49 per cent and a lack of expertise at 48 per cent. The flatness is the finding. There is no single dominant obstacle to clear, which means money, law and skills would each have to move for adoption to deepen, and a policy that addresses one of them leaves the other two in place.

The Commission points to the General Data Protection Regulation and the AI Act as rules that already address some of this, and says the forthcoming Quality Jobs Act "will increase clarity for companies and better protect workers' rights in relation to algorithmic management". EU social partners have already filed views in the second-phase consultation for that act. Whether the legislation reduces the 49 per cent who cite legal uncertainty or adds to it is, at this stage, unknowable.

What the survey cannot settle

Fieldwork closed in April 2026, six months before the Commission released the results, so none of the model releases of the past half year are reflected in the answers. The survey also asks firms what they do, not what it did to their output or their headcount, which is the question the evidence keeps declining to answer. Parallax Nexus has reported on Anders Humlum and Emilie Vestergaard's study linking Statistics Denmark surveys to administrative earnings records, which found no effect of AI chatbots on pay or hours across 11 highly exposed occupations.

There is one comparison worth keeping. The Digital Decade programme sets a target that at least 75 per cent of EU businesses take up AI, cloud or big data by 2030. On the adoption measure in this survey, 42 per cent now and 21 per cent intending, the bloc reaches 63 per cent only if every firm that says it plans to adopt actually does.