Data study ·
Companies most committed to AI hire up to 25% fewer young people on permanent contracts
Original title, in French: “Les entreprises les plus converties à l'IA embauchent jusqu'à 25 % de jeunes en CDI en moins”
- Publication
- ActuIA (France)
- Author
- Stephane Nachez
- Published on
- 20 October 2026 · read in the page structured data · updated on 20 October 2026
Summary
A research note by researchers from HEC Paris, Banque de France and INSEAD shows that in France, companies reporting intensive use of generative AI (around 600 of the 6,000 surveyed) hired 17 to 25% fewer 21-29-year-olds on permanent contracts (CDI) in 2025 than other companies in their industry, with no clear difference in total headcount. Within the same companies, the jobs most exposed to AI hire about one third fewer people than other jobs. Another analysis, based on a job exposure index built by Claude Opus 4.7 agents, puts the number of missing permanent contracts for 21-29-year-olds over one year at 75,000. The authors partially rule out other explanations (economic conditions, interest rates, remote working) and compare their results with similar studies in Denmark, the United States and Norway. They recommend redefining entry-level roles and preserving training pathways.
Summary based on the full text of the article.
Read the article on ActuIA →Figures cited
-
21 %
Gap in permanent-contract hiring of young people (21-29) at intensive AI-using companies compared with other companies in their industry, in 2025
France, private sector, ~600 intensive AI-using companies, 2025
« Selon les trimestres de 2025, ces entreprises recrutent en CDI 17 à 25 % de 21-29 ans de moins que les autres, soit un écart de -0,19 à -0,29 point de logarithme. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
600 companies
Number of companies reporting intensive use of AI among the companies surveyed
France, Banque de France monthly business survey, around 6,000 companies
« Environ 600 déclarent un usage intensif de l'IA. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
-33 %
Hiring gap between exposed and non-exposed jobs within the same intensive AI-using company, first half of 2025
France, 541 intensive users matched with Dares data, first half of 2025
« Au premier semestre 2025, les métiers les plus exposés y embauchent environ un tiers de moins que les autres métiers de la même entreprise. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
75,000 permanent contracts (CDI)
Missing permanent-contract hires over one year for 21-29-year-olds, compared with the November 2022 level
France, 21-29-year-olds, most exposed jobs, over one year to March 2026
« « Sur un an, cela représente environ 75 000 CDI de moins pour les seuls 21-29 ans », écrit la note. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
28.3 %
Decline in permanent-contract hiring of 21-29-year-olds in the quintile of jobs most exposed to generative AI, compared with November 2022
France, most exposed quintile, March 2026 vs November 2022
« Chez les 21-29 ans, les embauches en CDI du quintile le plus exposé sont en mars 2026 à 28,3 % sous leur niveau de novembre 2022, contre 7,5 % dans le moins exposé. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
42 %
Decline in permanent-contract hiring of developers and IT engineers aged 21-29, compared with November 2022
France, developers and IT engineers aged 21-29
« Chez les développeurs et ingénieurs informatiques de 21-29 ans, les embauches en CDI sont environ 42 % sous leur niveau de novembre 2022, contre 11 % dans les quatre quintiles moins exposés. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
10 %
Share of apprentices among 21-29-year-olds in the quintile most exposed to AI, end of 2023
France, 21-29-year-olds in the most exposed quintile, end of 2023 (vs 6% at end of 2019)
« Chez les 21-29 ans du quintile le plus exposé, la part d'apprentis passe de 6 % fin 2019 à 10 % fin 2023, selon les calculs des auteurs. »
Source: Note LIFT 2026-02, HEC Paris, Banque de France, INSEAD
-
9 %
Share of juniors in Anthropic's technical hires
Anthropic, internal data published on 16 October 2026, period not specified (vs 27% previously)
« Chez Anthropic, la part des juniors dans les embauches techniques est passée de 27 % à 9 %, selon des données internes publiées le 16 octobre. »
Source: Anthropic (données internes)
-
-59.5 %
Drop in the number of developer job postings published by Apec between 2022 and 2025
France, job postings published on the Apec website, 2022 to 2025 (31,330 in 2022, 12,690 in 2025)
« soit -59,5 % en trois ans (calcul d'ActuIA) »
Source: Apec, Les métiers cadres porteurs, édition 2026
-
-11 %
Decline in employment at Danish companies that adopted AI in 2023 or early 2024, compared with the 2023 trend
Denmark, companies that adopted AI in 2023 or early 2024, from 2023 to mid-2025
« Emploi de ces entreprises environ 11 % sous la tendance de 2023 à la mi-2025, recul concentré chez les jeunes et les plus diplômés »
Source: Danmarks Nationalbank, analyse n° 16
-
-19 %
Employment gap for 22-25-year-olds in the most exposed jobs in the United States, compared with their less exposed peers
United States, ADP payroll data to June 2026, 22-25-year-olds
« 22-25 ans des métiers les plus exposés : emploi 19 % sous la trajectoire des pairs moins exposés »
Source: Stanford Digital Economy Lab, Canaries in the Coal Mine?
Classification
- Country
- France, Denmark, United States, Norway
- Company size
- All sizes
- Industries
- All industries, or industry not specified · Software vendors, IT services companies, telecoms
- Companies
- HEC Paris (study author), APEC (study author), Numeum (cited), Dares (cited), Banque de France (study author), Anthropic (cited), Stanford Digital Economy Lab (study author), France Travail (cited), Danmarks Nationalbank (study author), INSEAD (study author)
- Keywords
- youth employment, recruitment, permanent contract hiring, generative AI, apprenticeships, developers
Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to ActuIA and its author. Entry translated from French by Claude Sonnet 5.5.