AI adopters hired less in Denmark. That is a warning, not an economy-wide layoff verdict.

What this study measured

Bonin, Darougheh and Kuchler match Danish firm-level AI-adoption information with monthly employer–employee records. They follow firms that first reported AI use in 2023 and compare their employment path with firms that did not report AI use, while accounting for prior trends and industry differences.

This is much closer to an adoption measure than an occupational exposure score. It still does not show which tool changed which task, or what would have happened if the same firm had not adopted AI.

The warning is in recruitment

By late 2025, the adopted firms' employment was about 11% below the path implied by their earlier trend relative to non-adopters. The result is concentrated in smaller firms and reflects lower recruitment, particularly in AI-exposed occupations. The official analysis says aggregate Danish employment had not shown a major change.

That makes the result consequential for a person watching new openings: a firm can expand more slowly because it hires fewer people without producing a visible wave of layoffs.

Why it is not a causal verdict on your job

Firms do not adopt AI at random. Adopters can differ in productivity, size, management or plans before they report using it. Following pre-adoption trends improves the comparison but cannot remove every difference. The paper covers Denmark, its firms and the period through late 2025—not the United States or every industry.

A current unemployment check points elsewhere

California's August AI-Unemployment Tracker reported a modest decrease in new initial unemployment claims from high-exposure occupations: 1.2% on its potential-exposure measure and 1.0% on its observed-use measure, both within recent fluctuations. That tracker sorts claims by occupational exposure; it does not identify firms that adopted AI or count workers who did not file for unemployment.

The two sources do not cancel each other. One follows hiring at firms that reported adoption in Denmark. The other tracks claims by exposure in California. Their contrast shows why “AI is cutting jobs” is still too blunt to guide a career decision.

What changes in the cumulative answer

The evidence now contains a sharper hiring warning: actual adopters in one country hired less, especially at smaller firms. It does not show an economy-wide employment fall, establish that AI alone caused the gap, or calculate your odds. Before acting on a headline, ask whether it measured reported adoption, openings, hires, unemployment or exposure—and where.