The labor market's safest workers are now its most exposed

Tenure no longer buys the same safety in the US labor market. The cohort that has historically found work fastest, primary workers with steady histories and long tenures, is now posting the steepest decline in job-finding rates of any segment, even while the broader economy keeps growing. The pattern is sorting, not cycling: the labor market is re-pricing by how much of a job's tasks a model can already do, rather than by workforce attachment.

Business Insider's reporting on a new Federal Reserve Bank of Richmond brief puts a number on the shift. From a November 2022 high to a September 2025 low, job-finding rates for primary workers fell 13 percentage points, against 2 points for secondary workers, the largest gap in the analysis. Before ChatGPT reached the public in late 2022, those rates moved together across AI-exposure levels. They have since diverged, with the steepest falls tracking the overlap between a role's tasks and the capabilities listed in AI patents.

The mechanism is portable to the next layoff cycle and the next hiring decision. A worker whose tasks overlap with what generative AI already handles at the task level faces a thinner on-ramp, regardless of resume. The labor market is no longer rewarding loyalty to a job; it is rewarding distance from a model.