AI in the Workplace: Record Adoption in France, Productivity Gains Still Limited in 2026

A new bulletin from the Bank of France confirms with hard numbers what many companies had already sensed for several months: generative artificial intelligence has spread through the French economy at remarkable speed, but it has not yet transformed productivity to the extent expected. Published on September 3, 2026, in the institution's Bulletin No. 266, the study offers a data-driven snapshot of AI adoption among sizable French companies, caught between genuine enthusiasm and economic effects that remain unconfirmed.

Adoption Crosses a Symbolic Threshold

According to data collected in early 2026, more than two-thirds of companies with at least 20 employees now report using generative AI tools, whether consumer-grade solutions such as conversational agents or features built directly into their business software. The Bank of France attributes this breakthrough to a particularly low barrier to entry: standardized tools require neither advanced technical skills nor specific data infrastructure, and employees can adopt them directly, sometimes even without formal approval from their IT department.

The Productivity Paradox

This is where the picture becomes more nuanced. Only 31% of companies using the technology say they have already observed a tangible productivity gain linked to generative AI. The Bank of France attributes this unevenness to uses that remain largely experimental, closer to individual trial and error than to any organized overhaul of business processes. This gap between mass adoption and limited economic effects is not entirely new: agentlink.org had already discussed the structural digital productivity gap between France and the United States, a gap that AI could eventually close, but clearly has not closed yet.

Where AI Is Truly Taking Hold Inside Companies

The bulletin notes that usage remains concentrated in support and cross-functional roles—administration, internal communication, document drafting, and customer relations—rather than in the core of production processes. This partial integration echoes a pattern already documented elsewhere: several major corporations have seen their first AI agent pilots stall for lack of a clear method, to the point that some consulting firms are now creating dedicated units to avoid the large-scale failure of these experiments. In addition, companies deploying AI systems must contend with an increasingly demanding regulatory framework, including the new transparency requirements under Article 50 of the European AI Act, which add a layer of compliance that must be built in from the deployment phase onward.

The Labor Market, the First Proving Ground

Beyond the macroeconomic figures, it is also in the labor market that the effects of this rapid spread are being felt first. agentlink.org recently detailed how the arrival of AI is reshaping the job market for young graduates, who are often on the front line of tasks that can now be partially automated. The Bank of France also notes that the strong variation in usage across sectors and company sizes points to very uneven transformation trajectories in the coming years.

What Lies Ahead?

For the institution's economists, the current gap between adoption rates and measured gains is nothing unusual at this stage: general-purpose technologies have historically taken several years to produce visible effects on overall productivity, as organizations need time to reorganize their processes around the tool rather than simply layering it on top of existing workflows. The shift toward deeper integration, within the actual value-creation chains rather than just peripheral functions, will likely be the next indicator to watch in order to measure whether AI delivers on its economic promise.

Frequently Asked Questions

How many French companies use generative AI in 2026?

According to the Bank of France, more than two-thirds of companies with at least 20 employees report using generative AI tools in early 2026, a level of adoption the institution describes as rapid.

Is generative AI already improving companies' productivity?

Not yet on a widespread basis: only 31% of companies using the technology report a concrete productivity gain, with usage remaining largely experimental and concentrated in support functions.

Why is AI adoption outpacing its measurable effects?

The Bank of France attributes this gap to a low barrier to entry, which encourages spontaneous adoption by employees, as well as to the still-imperfect integration of these tools into companies' production processes.

Sources

• Bank of France — Bulletin No. 266
• Boursorama
• Batiactu

For more, check out our coverage in the Artificial Intelligence section and the Economy section of agentlink.org.

This article was written with the help of artificial intelligence. Editorial policy