French junior hiring fell fastest in AI-exposed work

What changed

The authors use French private-sector administrative records on hires and contract endings from January 2019 to March 2026. Permanent-contract endings among 21–29-year-olds also fell: 23% in the most-exposed quintile versus 6% in the least-exposed. Hiring fell faster than departures. This points to a slower entry route, not a wave of measured layoffs.

About 6,000 firms were linked to a Banque de France survey of AI-use intensity fielded once in early 2026. Firms reporting intensive use also had relatively weaker junior permanent hiring, but their total headcount did not clearly diverge. The firm survey strengthens the relevance of adoption; it does not remove selection or prove causation.

Anthropic’s internal study gives a separate one-firm case: junior roles went from 27% to 9% of its technical hires since early 2025, while the company expanded. Technical hiring still grew in absolute terms. This is a changing share at an unusually AI-intensive firm, not a national job-loss rate.

What the studies cannot answer

The French exposure score estimates which occupational skills AI could perform. It does not measure each employer’s use. Post-pandemic tech hiring corrections, remote work, demand and apprenticeship changes could contribute to the gap. The administrative flow series does not establish the number of young people employed or unemployed in each occupation. Anthropic’s report uses its own internal records and does not separate AI from rapid company growth.

These findings sharpen the warning about entry-level access. They do not overturn Yale’s broad U.S. no-clear-footprint check, which measures a different population and outcome.