Unemployment data for recent US college graduates in the summer of 2026 shows rates within historical norms, according to a CESifo study. This contrasts with some earlier predictions and studies suggesting AI would significantly impact new graduates' job prospects.

The findings challenge widespread concerns that AI adoption would immediately lead to higher unemployment for new college graduates, suggesting that the labor market is currently absorbing these technological shifts without unusual disruption for this demographic.
The US job market for recent college graduates in the summer of 2026 appears remarkably normal, with an unemployment rate of 7.3% that falls within the range seen in previous years, according to research from CESifo. This finding contrasts with some earlier predictions and studies, such as one from Stanford based on ADP payroll data, which suggested a more significant negative impact of AI on new graduates' employment prospects.
The CESifo researchers conducted several comparisons, including against non-college graduates and older college graduates, and analyzed employment by potential AI exposure. In nearly all comparisons, any trend differences observed between 2022 and 2026 were not statistically significant. The data consistently show that unemployment among recent college graduates in the summer of 2026 was not unusually high relative to earlier summers.
The study attributes the difference in findings partly to the data sources used. The CESifo research relied on a wider Census survey, while the Stanford study used ADP payroll data, which may not capture all economic elements. Furthermore, the CESifo analysis considers both job supply and aggregate demand, noting that demand could shift even if AI contracts job supply in certain fields.
While the current data suggest AI has not yet significantly impacted recent graduates' unemployment, the CESifo researchers caution that this may change. They warn that if AI use in the workplace continues to intensify, graduating classes from 2027 onward could face greater effects, and more years of data will be necessary to confirm emerging trends.
Pick the topics you care about. Get only what matters, on your cadence.