Skip to content
Marsanne Studio
← Back to home
← The AI watch

Weekly digest ·

AI weighs on youth employment in France as the gap between SMEs and large groups persists

A study by the Banque de France and HEC confirms a sharp fall in permanent-contract hiring of young people in the occupations most exposed to generative AI. Across markets, the adoption gap between large companies and SMEs remains clear, despite new public schemes in France and Spain. In the United States, companies that are further ahead on agentic AI report better returns, but the governance of agents is lagging behind their deployment.

Key developments

  1. Permanent-contract hiring of young people falls in AI-exposed occupations

    A study by the Banque de France and HEC Paris's LIFT shows that permanent-contract (CDI) hiring of 21-29 year-olds fell by 28% between November 2022 and March 2026 in the occupations most exposed to generative AI, against 7.5% to 8% elsewhere, or around 75,000 fewer permanent hires a year. The IT industry is the hardest hit, with a 42% drop in hiring of young developers. In consulting and engineering, Capgemini, CGI and Accenture have reduced their headcount in France, and the Opiiec barometer reports that 7% of companies in the Syntec sector have already cut staff on automatable tasks.

    Sources: Le Monde Informatique, L'Usine Nouvelle, Alliancy, Siècle Digital

  2. SMEs and large companies remain unequal in the face of AI, in France and in Spain

    In France, a Banque de France study shows that more than two thirds of companies with over 20 employees had adopted generative AI by early 2026, but only 30% of SMEs with fewer than 50 employees use complex AI, against 60% of large companies. In Spain, the government has announced a €600 million bono for autónomos and SMEs under the IA360 Plan, with a target of 55% of companies using AI in 2030, against 21.1% currently. In the Basque Country, adoption stands at 47.5% among large companies against 8.7% among small businesses.

    Sources: Culture RH, Diario Autónomos y Emprendedor, SPRI, Cadena SER, Cinco Días, Cadena de Suministro

  3. AI agents are being deployed faster than they are governed

    Several studies point to a gap between the deployment of AI agents and the controls around them. According to the 2026 report by IBM and the Ponemon Institute, 92% of organisations that suffered an AI-related breach lacked adequate access controls. A study cited by Journal du Net reports, citing McKinsey, that more than 60% of organisations are experimenting with AI agents, of which 97% are said to have excessive privileges. The IFS Industrial Index 2026, based on more than 2,200 industrial executives in 14 countries, shows that 43% of operational teams have integrated AI agents within a year, but only 11% validate their results on critical processes.

    Sources: Journal du Net (1, 2), Solutions Numériques, LeMagIT, Industria Química

  4. Nvidia invests in the French AI ecosystem, led by Doctolib

    Nvidia is taking a minority stake in Doctolib, as well as in three other French companies (Pigment, Genesis AI, Entalpic), as part of a European strategy in which the group says it has invested nearly $10 billion on the continent. Doctolib, which says it serves 520,000 healthcare professionals and 90 million patients, wants to accelerate the development of proprietary AI models. The deal is part of a European tour by Jensen Huang taking in Paris, London, Stockholm and Berlin, and comes as computing infrastructure is scaling up in Europe, including the Scaleway supercomputer in Sweden.

    Sources: Le Figaro, FrenchWeb (1, 2), Journal du Net, Challenges, Le Monde Informatique

  5. Accounting and law: AI is redefining how production is shared

    Online accounting firm Dougs has raised €50 million to strengthen AI in the processing of accounting files, in a French market that is consolidating. According to KPMG, 75% of companies now use AI in their finance function, up from 30% in 2024, but only 42% have auditable AI. In law, a Karnov survey conducted in five European countries shows that 31% of legal professionals use AI daily to draft documents, while in France large consulting and engineering firms are launching headcount reduction plans linked in part to automation.

    Sources: FrenchWeb (1, 2), Les Echos, Cinco Días, Compta Online

Figures of the week

  • 75,000 permanent hires/year

    Annual fall in permanent-contract hiring of 21-29 year-olds in the occupations most exposed to generative AI

    France, 21-29 year-olds, occupations most exposed to generative AI, since November 2022

    « ces professions enregistrent environ 75 000 embauches en CDI de moins par an chez les 21-29 ans, selon une étude de la Banque de France et du Laboratoire de l’innovation et du futur du travail (LIFT) d’HEC Paris. »

    Le Monde Informatique , from Banque de France et LIFT HEC Paris, étude « IA générative : les premiers changements du marché du travail sur les jeunes »

  • 30 %

    Share of SMEs with fewer than 50 employees using complex AI

    French SMEs with fewer than 50 employees

    « Seules 30 % des PME de moins de 50 salariés y ont recours, contre 60 % dans les grandes entreprises. »

    Culture RH , from Banque de France

  • 600 million EUR

    Amount of the bono planned for AI adoption by autónomos and SMEs

    Spain, IA360 Plan

    « El Gobierno ha anunciado un bono de 600 millones de euros para que autónomos y pymes incorporen inteligencia artificial a sus negocios. »

    Diario Autónomos y Emprendedor , from Plan IA360 (Gouvernement espagnol)

  • 80 %

    Share of managers and executives reporting positive returns from their department's AI investments

    Managers and senior managers at US private companies with 1,000+ employees, July 2026

    « El 80 % de los gerentes y los encuestados de cargos superiores afirma que las inversiones de su departamento en IA generan retornos positivos, pero solo el 28 % los califica de muy positivos. »

    Unite.AI , from Wharton Human-AI Research / GBK Collective, « La ventaja adaptativa : quién está ganando con la IA y cómo »

  • 75 %

    share of employees using AI regularly

    global, 11,749 workers surveyed in 14 countries

    « les trois quarts des salariés utilisent désormais régulièrement l’IA »

    IT for Business , from Boston Consulting Group, quatrième enquête mondiale annuelle

In France and Spain

France

Beyond the fall in youth hiring and Nvidia's investment, the week confirms the spread of AI into concrete business uses. Public hospitals are rolling out tools to reduce doctors' administrative workload, while Spie and other groups are testing AI agents for their security operations centres (SOCs), with verdict validation rates above 90%. The MCP protocol makes it easier to connect AI assistants to SME management software, raising new governance questions. Public schemes also target micro-businesses and SMEs: the France Legaltech programme, the Plan d'Amplification Performance IA for fashion, and training by professional branches.

Sources: Journal du Net, LeMagIT, IT for Business (1, 2, 3), Alliancy, Direction générale des Entreprises, France Num, ActuIA

Spain

Besides Spain's public plan for AI adoption and its €600 million bono, several Spanish studies detail the gap between individual use and organisational integration of AI. According to EY and enClave de Personas, 88% of employees already use AI and save eight hours a week on average, but only 5% reach an advanced level of integration. Banco Sabadell is preparing a large-scale rollout of AI agents with T-Systems for 2027. In healthcare, the BIAlert Sepsis tool brought sepsis mortality down from 13% to 6% at a hospital in Palma de Mallorca. The Cámara de Comercio points out that only 13% of Spanish SMEs already operate with AI.

Sources: RRHH Press, Expansión, RRHH Digital, Revista Byte TI, Cinco Días (1, 2), SPRI

Internationally

United States

A study by Wharton Human-AI Research and GBK Collective, based on 1,416 employees of US companies, shows that 80% of managers see positive returns from AI, but only 28% consider them very positive; organisations that have created a chief AI officer role or standardised on a single platform achieve markedly better results, and 37% of managers report AI-related headcount reductions. Several large companies are tightening control of AI agents: Citi is restricting their access to bank accounts, American Express is setting rules for their purchases, and many retailers are blocking shopping agents such as Meta's Muse. Epic Games is making a full shift to AI-generated code after 1,000 job cuts, while Getty Images has filed for Chapter 11 protection, partly as a result of generative AI.

Sources: Unite.AI (1, 2), Financial Times, TechCrunch, Xataka (1, 2, 3)

China

China features this week mainly through the adoption of its open models abroad: the German search engine Ecosia has dropped Mistral in favour of open-source Chinese models (Qwen, GLM, Kimi), citing a quality gap and the French provider's lack of independence. The article notes that 70% of the electricity powering Chinese data centres came from coal in 2025, with a target of 80% renewables by 2030.

Sources: Xataka

Rest of the world

A global Boston Consulting Group survey of 11,749 workers in 14 countries shows that three quarters of employees now use AI regularly, with rates reaching 95% in India, but that half of users spend more time checking AI than working directly. In Singapore, Anthropic has signed a memorandum of understanding with the government and says it has deployments at GovTech, DBS and Grab. In Chile, a programme aims to train 150,000 SMEs with Google and Amazon. In Mexico, a Strand Partners study for AWS puts the share of companies already using some form of AI at 48%, mostly for basic uses.

Sources: IT for Business, Unite.AI (1, 2), El Ecosistema Startup, Mexico Business News

What's changing

Over the four comparison weeks (W38 to W41), the share of “all sizes” articles rises to 69% from 63% the previous week, while the share devoted to large companies falls to 26% from 31%. The consulting and legal industry rises from 7% to 13% of articles, driven by studies on consulting and engineering firms. The keyword “cybersécurité” enters the top themes for the first time (8%), having been less prominent in previous weeks, while the share of articles mentioning “pme” falls from 13% to 9%.

Comparison with the previous 4 weeks

Method

Digest written on 19 October 2026 by an AI model (Claude Sonnet 5, by Anthropic) from the 198 articles published in the AI watch for 12 – 18 October 2026. Each paragraph links to the entries it draws on, and every number in the text was found in one of those articles before publication. The figures of the week are given with the exact sentence from the article that provides them. Digest translated from French by Claude Sonnet 5.5.

Previous weeks