OpenAI published a new report today mapping AI's likely near-term impact on jobs across the European Union, country by country. The report extends a framework OpenAI first built for the US labor market in April 2026, now applied to European occupations using official EU employment data.
The framework puts every occupation into one of four categories. About 12% of EU employment is in jobs that may actually grow as AI lowers costs and makes more projects viable. About 14% sits in occupations with relatively high near-term automation potential. Another 27% covers jobs where AI changes the way work is done without removing people entirely. The remaining 47% sees little immediate change.
Those four categories are not forecasts. They are a planning map, a tool for asking better questions about where pressure is building before it shows up in unemployment statistics.
The country picture matters as much as the EU-wide numbers. Germany, Greece, and Italy have larger shares of workers in the higher-automation-risk group, which reflects the structure of those economies. Germany has a large manufacturing and industrial base. Italy and Greece have significant concentrations of routine service and administrative roles. Sweden, Luxembourg, and the Netherlands, with stronger knowledge-economy foundations, have more workers in the "may grow" category.
This is consistent with what independent researchers have found. The German Institute for Employment Research projected that 1.6 million jobs in Germany alone could be reshaped or lost to AI over the next fifteen years. The European Central Bank noted in March 2026 that there is currently little evidence of a major employment hit in the euro area, but was clear that displacement risks remain real over the longer term.
The more important observation, confirmed by researchers at Carnegie Endowment and the European Policy Centre, is that the change will not look like sudden mass redundancy. It will look like jobs that slowly get narrower, with fewer tasks, less responsibility, and weaker pay. Workers in those roles will feel insecure well before official data registers anything alarming.
For business operators specifically, two things stand out. First, the AI adoption gap between large companies and small ones is real and widening. In 2024, 41% of large European enterprises were using AI; only 11% of small companies were doing the same. European SMEs make up 99% of all EU businesses and employ nearly half the workforce, so that gap matters for the broader economy, not just for individual firms. Second, even the companies that are experimenting with AI are mostly using it on peripheral tasks: writing, summarizing, and support functions. Only 29% of SMEs using AI apply it to their core activities, which means productivity gains at the ground level are still modest.
For operators managing teams in back-office, administrative, or data-heavy roles, the OpenAI report effectively confirms what many already sense: workflows in those areas are going to change faster than HR planning cycles typically account for. The time to think about where staff roles are heading is before job postings reflect the shift, not after.
OpenAI's interest in publishing this kind of research is not neutral. The company is making a policy and public-relations argument, as well as expanding its footprint in European government conversations. That does not make the analysis wrong, but it is worth reading it alongside independent EU and ECB research, which largely agrees on the direction even if it is more cautious about the speed.