France finds itself at a strange intersection of the global AI arms race accelerates where the traffic lights are flashing green, orange, and red all at once.
On one hand, the government wants to hit the gas. On the other hand, regulators are slamming the brakes. And somewhere in the middle, a growing number of VCs and founders are asking Brussels to stall the European AI Act, the first measures of which came into effect just before France’s AI Act in February.
Even the French public seems confused.
An IPSOS survey published in January found that 73% of French citizens fear AI’s impact on society. That's the highest level of concern in a global panel of 11 countries. And while 48% of salaried workers worry AI will replace their job (BVA, 2024), only 10% of French companies used AI at all in 2024 – a mere 4 point increase over 2023, according to the latest data from INSEE, the nation's statistics agency.
That’s well behind neighbors in Northern Europe, where adoption sits between 20–28%, and lower than the EU average of 13%.
So, what’s really going on? Is France racing toward an AI-powered future or hesitating at the crossroads?
🟢 Osez l’IA: Green Light for Adoption
Last week, Clara Chappaz, Minister Delegate for AI and the Digital Economy, unveiled “Osez l’IA", a €200 million national plan designed to bring AI into the mainstream economy. The launch, hosted at Station F, sent a clear message: for France’s businesses, AI is no longer optional.
“We have the brains, but not the culture," Chappaz said. "We have the skills, but not yet the usage. We have the innovation, but not the trust...Osez l'IA is our response to these questions, these concerns. The ambition is to bring artificial intelligence into the daily lives of all our companies.”
By 2030, the plan aims to have 100% of large enterprises, 80% of all SMEs, and mid-caps and 50% of micro-enterprises all actively using AI.
To do so, the “Osez l’IA” program will roll out a number of measures built around three major strategic pillars:
1. Diagnose: 300 “AI Ambassadors” drawn from Bpifrance, FranceNum, CCI, French Tech, and others will be deployed across France to help businesses identify how AI fits their operations. They’ll host monthly matchmaking events at Bpifrance and CCI offices across France, as well as an annual AI Business Day,the pilot edition of which took place last year. In the meantime, public funding will subsidise up to 40% of the cost of AI-readiness diagnostics.
2. Train: By year-end 2025, the government will launch the AI Academy, an open-access platform offering tailored courses and tutorials for all: students, job seekers, employees, and business leaders alike. Through programs like France 2030’s “Compétences et métiers d’avenir,” the aim is to train 15 million professionals in AI by 2030.
3. Support: Once diagnosed and trained, companies will have access to:
- A catalogue of off-the-shelf AI tools, grouped by industry and use case
- State-guaranteed loans via Bpifrance for implementation
- An 18-month acceleration programme for high-impact firms
- Expanded public-private research partnerships, including 10 new SME collaborations per year through INRIA.
A national Citizen consultation, open until August 31st via Make.org, also invites the public to weigh in on AI’s role in society, reflecting a broader effort to engage all stakeholders, not just industry insiders.
"This plan also aims to strengthen our digital sovereignty by highlighting French AI solutions," Chappaz said. "I am here to launch a dynamic. But it is thanks to you that AI will spread.”
🛑 Red Lights from the Regulators
While the government is cheering AI adoption, regulators are urging caution.
In its 2025–2028 strategic plan, France’s data watchdog CNIL outlines a growing list of concerns centered around AI and its increasing adoption both in the workplace and the home, including algorithmic bias, AI-driven misinformation, risks to children’s privacy, and businesses' increasing vulnerability to cyberattacks.
The CNIL’s position? Let’s prepare for AI’s risks while enabling its benefits. This includes clear legal frameworks, system audits, and education campaigns with special attention on generative AI and the protection of minors.
Elsewhere, France’s telecoms regulator, Arcep, published a striking contribution to the European Commission in July, warning that Europe risks repeating its “cloud sovereignty failure” if it doesn’t act faster on AI infrastructure and regulation.
Drawing lessons from France’s fiber rollout - a case study in successful regulation - Arcep argues that the EU needs to simultaneously ensure competitive and open cloud/AI markets, prevent lock-in from Big Tech companies, all while supporting sustainable digital infrastructure (AI already consumes ~10% of France’s electricity).
If Arcep doesn’t reference the European AI Act directly, their call for “swift, harmonized, and enforceable” oversight is hard to mistake.
Even the French Senate weighed in this spring. In a May 2025 opinion, the European Affairs Committee criticized the AI Act for being “not ambitious enough,” particularly regarding copyright and general-purpose AI models.
🟠 Or Just… Pause?
Amid all the acceleration and alarm, some are now asking for a timeout.
In late June, inspired by the Swedish prime minister’s labeling of the EU’s new AI rules as “confusing,” nearly 50 European business and tech leaders – including French companies like Airbus, BNP Paribas, Partech and Pigment – signed an open letter urging Brussels to pause implementation of the AI Act…for two years. Their complaint: the regulation, particularly the Code of Practice for General-Purpose AI Models, is fragmented, unclear, and too early-stage for such a fast-moving field.
“Startups pivot all the time. European regulation needs to do the same,” said Partech General Partner Reza Malekzadeh. He believes that innovation should come before regulation and that “we need to pause, adapt, and regulate with today’s reality in mind.” He added, “I’m not saying we shouldn’t do it, I’m saying it's not the right time.”
Pigment CEO Eléonore Crespo agrees. She also signed the open letter stating on LinkedIn that, “The AI Act comes from a good place. But the way it’s being rolled out could seriously hurt the very startups building the future.”
Their warning echoes memories of GDPR: a landmark regulation, but one that many say disadvantaged startups while larger tech giants absorbed the compliance costs.
For its part, the EU says: No dice. "There is no stopping the clock, there is no grace period, there is no pause," a Commission spokesperson said Friday, according to Euractiv.
🚥 The French Paradox
France today is the epitome of AI ambivalence. The government is investing hundreds of millions to promote AI across every sector, while the public remains deeply uneasy, regulators warn of unchecked dominance and environmental tolls. And industry leaders plead for more clarity - or just more time.
The message would appear, in short, to be, “Accelerate, but regulate."
