07/04/2026
Safety for the future
A New Foresight Workshop on AI and Safety
On April 2, Foncsi organized a new workshop in Paris with its patrons and partners on “The use of AI in companies: what positive and negative safety-related challenges are associated with it?” A look back at an afternoon rich in questions and discussions.
Building on the success of a first edition held in 2025 on the safety of the future in the face of major societal, environmental, and industrial changes, the Foundation has just organized a new foresight workshop on AI and safety, on April 2, 2026, at the Maison de la Recherche in Paris, with around thirty of its patrons and partners, representing industrial companies, regulatory authorities, trade union organizations, and others.
• Airbus
• Anssi
• ASNR
• CFE-CGC
• CGT
• DGAC-DSAC
• EDF
• Engie
• EPSF
• Icsi
• IMDR
• Ineris
• Natran
• OPPBTP
• RTE
• SNCF
• Suez
• Systra
• TotalEnergies
• UIC
.
A new Model: the “AI and Safety Daisy”
The Foncsi team, and more specifically its director René Amalberti, unveiled a mapping of the promises of AI for safety and the associated risks:
PROMISES of AI & the associated risks
Anticipated risk | The “push for innovation” and the retreat from a precautionary strategy
Reduced hazard exposure | Devaluation of expertise; disruption of professions and jobs
Protected people | The end of managed work: constrained people, loss of autonomy and meaning
Support for decision-making | Complacent people unable to exercise judgment
A monitored and secured system | AI‑based malicious use, a cyber risk exploding in its diversity
Enhanced feedback and learning | Feedback and learning that is too rich and blind beyond data to be usable
Promises of AI & Safety
Risks of AI & Safety
Based on the “AI and Safety Daisies,” participants worked in breakout groups to identify what they could prioritize deploying within their organizations to achieve rapid improvements in safety.
The arrival of AI: what trajectory?
A second plenary session provided an opportunity to address the different scenarios of risk evolution that have already occurred during previous major industrial revolutions: the introduction of aircraft automation (1980–2000) and the advent of digital technologies 2.0 and then 3.0 (1990–2020).
The different groups then explored whether it was possible to draw inspiration from these evolutionary trajectories (faster pace? what type of curve? rebounds toward other risks? societal consequences? etc.) and presented proposed risk evolution curves associated with the arrival of AI.
Coming up...
This afternoon of exchanges was particularly rich. Participants were able to explain the projects already under development and what they found especially promising in AI for improving industrial safety. They also exchanged views on their concerns: loss of skills, the role of humans, performance, data reliability, and more.
A synthesis of these workshops is currently being written and will be widely disseminated.
Finally, the day was filmed, with a video highlighting key moments to be released soon.
Note: the organization of the next foresight workshop is already being considered for October 2026 in Paris.
