Observatory of
Responsible AI
2026 edition
Artificial intelligence is advancing rapidly. Its applications are expanding just as fast.
Responsible practices, however, are struggling to keep pace.
Amid rapid adoption, limited training, and the absence of clear governance frameworks, how are organizations and their employees truly experiencing AI today?
The 2026 edition of the Responsible AI in Business Observatory explores the uses, expectations, and tensions emerging as AI becomes increasingly embedded across the workplace.
Observatory of
Responsible AI
2026 edition
Artificial intelligence is advancing rapidly. Its applications are expanding just as fast. Responsible practices, however, are struggling to keep pace.
Amid rapid adoption, limited training, and the absence of clear governance frameworks, how are organizations and their employees truly experiencing AI today?
The 2026 edition of the Responsible AI in Business Observatory explores the uses, expectations, and tensions emerging as AI becomes increasingly embedded across the workplace.
The Numbers That Matter
Adoption is accelerating. Support is struggling to keep pace.
90%
of employees trained in responsible AI are already using generative AI in the workplace, compared to …47 %
on average among private-sector employees in France.
23%
only report having received training in the responsible use of AI, of which …6%
have undergone comprehensive training on the subject.
61%
use external, consumer-grade AI solutions, a phenomenon commonly referred to as “Shadow AI,” compared to …19%
who rely on internal AI tools developed by their organization.
10%
report that their organization has an AI ethics charter or committee in place, while …54%
consider responsible AI to be desirable—or even essential.
The Numbers That Matter
Adoption is accelerating. Support is struggling to keep pace.
90%
of employees trained in responsible AI are already using generative AI in the workplace, compared to …47 %
on average among private-sector employees in France.
23%
only report having received training in the responsible use of AI, of which …6%
have undergone comprehensive training on the subject.
61%
use external, consumer-grade AI solutions, a phenomenon commonly referred to as “Shadow AI,” compared to …19%
who rely on internal AI tools developed by their organization.
10%
report that their organization has an AI ethics charter or committee in place, while …54%
consider responsible AI to be desirable—or even essential.
AI for everyone means saying clearly: yes, it’s allowed; yes, secure and governed tools are made available; yes, employees are trained before using them and on an ongoing basis to improve their practices while combining performance and responsibility. But being an augmented human doesn’t mean letting AI think in your place: it means keeping your judgment and analytical ability, using AI when it genuinely helps, and staying in control of the outcome. Responsibility, at its core, means asking the right questions from the outset about the impact of your practices, including in the long term.
Aldrick Zappellini, Chief Data & AI Officer Groupe - Crédit AgricoleThe more we communicate about what we’re doing with AI within the company, the more we foster a climate of trust. Training in the responsible use of AI is an absolute necessity: it must start from day one, then be adjusted continuously. And at every stage, employees must retain the choice between manual and automated approaches, so that they remain the final judges—not AI.
Karien Hervé, Executive Board Member, Groupe HervéAI isn’t a separate topic: it runs through governance, employability, and environmental impact. Training in responsible AI is both urgent and important: the urgency lies in not being left behind; the importance lies in learning to master AI rather than being swept along by the transformation. We can’t delegate everything to AI, especially not control over the information we put out. The final word must remain with humans, through the exercise of their professional judgment and critical thinking.
Thaël Boost Huard, Director of the Impact Program - KPMG FranceEurope and France have a major role to play in AI-based applications: we have global leaders in energy, water, waste, and machinery—and all of that expertise needs to be built into AI models. But being responsible with AI isn’t just about managing legal risk, confidentiality, or compliance: it’s also about training employees so that no one gets left behind.
Philippe Rambach, Chief AI Officer - Schneider ElectricThe shadow AI issue raises real questions: if the tool made available within the company isn’t suited to the need, employees will find other ways to get things done. That’s why AI governance needs to be the single point that captures every AI use case, to frame initiatives and prevent them from developing separately in silos. We’re still at the stage of building our 360° AI Strategy: that doesn’t stop us from testing, building awareness, and supporting employees in the responsible use of AI, but the challenge is to lay the groundwork for what we want for the future.
Charlotte Krieg, Innovation Director - Aésio MutuelleCompanies must start now to prepare for the jobs of tomorrow and organize the use of AI for the long term.
Jean-Manuel Soussan, Vice President and HR Director – BouyguesThe challenge of Shadow AI within organizations has become strategic.
Albane Liger-Belair, Innovation Director – KPMG FranceEmployees need to know what they can entrust to AI, what they need to verify, and what data must not be shared.
Roxana Rugina, Executive Director – Impact AIAI for everyone means saying clearly: yes, it’s allowed; yes, secure and governed tools are made available; yes, employees are trained before using them and on an ongoing basis to improve their practices while combining performance and responsibility. But being an augmented human doesn’t mean letting AI think in your place: it means keeping your judgment and analytical ability, using AI when it genuinely helps, and staying in control of the outcome. Responsibility, at its core, means asking the right questions from the outset about the impact of your practices, including in the long term.
Aldrick Zappellini, Chief Data & AI Officer Groupe - Crédit AgricoleThe more we communicate about what we’re doing with AI within the company, the more we foster a climate of trust. Training in the responsible use of AI is an absolute necessity: it must start from day one, then be adjusted continuously. And at every stage, employees must retain the choice between manual and automated approaches, so that they remain the final judges—not AI.
Karien Hervé, Executive Board Member, Groupe HervéAI isn’t a separate topic: it runs through governance, employability, and environmental impact. Training in responsible AI is both urgent and important: the urgency lies in not being left behind; the importance lies in learning to master AI rather than being swept along by the transformation. We can’t delegate everything to AI, especially not control over the information we put out. The final word must remain with humans, through the exercise of their professional judgment and critical thinking.
Thaël Boost Huard, Director of the Impact Program - KPMG FranceEurope and France have a major role to play in AI-based applications: we have global leaders in energy, water, waste, and machinery—and all of that expertise needs to be built into AI models. But being responsible with AI isn’t just about managing legal risk, confidentiality, or compliance: it’s also about training employees so that no one gets left behind.
Philippe Rambach, Chief AI Officer - Schneider ElectricThe shadow AI issue raises real questions: if the tool made available within the company isn’t suited to the need, employees will find other ways to get things done. That’s why AI governance needs to be the single point that captures every AI use case, to frame initiatives and prevent them from developing separately in silos. We’re still at the stage of building our 360° AI Strategy: that doesn’t stop us from testing, building awareness, and supporting employees in the responsible use of AI, but the challenge is to lay the groundwork for what we want for the future.
Charlotte Krieg, Innovation Director - Aésio MutuelleCompanies must start now to prepare for the jobs of tomorrow and organize the use of AI for the long term.
Jean-Manuel Soussan, Vice President and HR Director – BouyguesThe challenge of Shadow AI within organizations has become strategic.
Albane Liger-Belair, Innovation Director – KPMG FranceEmployees need to know what they can entrust to AI, what they need to verify, and what data must not be shared.
Roxana Rugina, Executive Director – Impact AIListen, watch, read
Read full report
Detailed analyses, key figures, and interviews with executives and decision-makers.
Shareable infographic
Key study findings in a ready-to-share visual format.
Express Audio
Key takeaways to listen to in a few short minutes—anytime, anywhere.
Make AI a lever for trust
and performance
Listen, watch, read
Read full report
Detailed analyses, key figures, and interviews with executives and decision-makers.
Shareable infographic
Key study findings in a ready-to-share visual format.
Express Audio
Key takeaways to listen to in a few short minutes—anytime, anywhere.
Conducted by
in partnership with
Conducted by the ViaVoice institute for Impact AI, in partnership with KPMG France and Bouygues for the 2026 edition, the Observatory of Responsible AI in Business has existed since 2018 and is now in its 6th edition. This year, it is based on an online survey of a representative sample of 1,000 private-sector employees in mainland France, aged 18 and over, surveyed from February 17 to 26, 2026. For the first time, the study was enriched with in-depth interviews conducted with executives and decision-makers from ten member organizations: Aésio Mutuelle, AXA, Bouygues, Crédit Agricole, Groupe Hervé, KPMG, Lesaffre, Orange, Salesforce, and Schneider Electric. Previous editions of the Observatory are also available on the Impact AI website.
Conducted by
in partnership with
Conducted by the ViaVoice institute for Impact AI, in partnership with KPMG France and Bouygues for the 2026 edition, the Observatory of Responsible AI in Business has existed since 2018 and is now in its 6th edition. This year, it is based on an online survey of a representative sample of 1,000 private-sector employees in mainland France, aged 18 and over, surveyed from February 17 to 26, 2026. For the first time, the study was enriched with in-depth interviews conducted with executives and decision-makers from ten member organizations: Aésio Mutuelle, AXA, Bouygues, Crédit Agricole, Groupe Hervé, KPMG, Lesaffre, Orange, Salesforce, and Schneider Electric. Previous editions of the Observatory are also available on the Impact AI website.
