Artificial intelligence is no longer a subject of innovation. It is a governance issue. Three recent cases illustrate how poorly managed AI opportunities can turn into reputational, legal or strategic crises when they could have been, on the contrary, Game Changers.
Deloitte: generative AI at the heart of a government scandal
The opportunity here is to become more productive in writing a report for the Australian government on a new public policy. The problem is that this report, which was partially written by IA, contained factual errors and fictional references. Government officials realized this with serious consequences: partial reimbursement of the contract, loss of credibility, regulatory tightening. The real reasons for the scandal are the lack of a rule on human validation and on declarations of use of AI. It is also unclear about the place of productivity in the search for performance.
Volkswagen: phantom braking linked to embedded AI systems
In the case of the manufacturer, the opportunity is the competitive advantage that innovation brings. The new advanced driver assistance systems integrate AI modules (perception, sensor fusion, obstacle detection, behavior learning) to improve passenger safety. The problem came from hundreds of cases of untimely braking linked, among other things, to poorly recalibrated sensors. Customer feedback led to a national investigation and an internal audit was initiated. The causes are linked to insufficient algorithmic governance: embedded AI requires continuous monitoring, which distinguishes it from classic computer bugs. Nobody, in the chain of governance, had alerted to the following fact: a windshield replacement or a minor collision can disrupt the AI system
Mediahuis : AI in ambush and damage to editorial credibility
For Mediahuis, as for most European press groups, AI represents a very rational opportunity to increase editorial productivity, to personalize information on its platforms and to experiment with new formats. The problem came from integrating AI-generated content into several of its newspapers without explicit mention. Readers discovered the use of AI through leaks and not through the transparency of the group. Result: accusations of deception, damage to editorial credibility, feeling of betrayal expressed by the editors. It is explained by a lack of challenge in the AI strategy: the technological subsidiary Mediahuis TPS had obtained the green light from the general management to fuel the publication of content driven by the journalistic work of the group's 30 editorial offices without them being consulted.
What these events say about the Councils:
Boards did not see how responsibility, risk, and value have evolved in an AI-dominated world. Historically, boards oversaw technology as a matter of execution: delegated, reversible, and limited in scope. AI breaks all three of these assumptions. Risk builds up outside of formal supervision. AI is squeezing organizations. AI is expanding the registers and the scope of risks. Where is the pilot in this new situation, if not at the highest level?
The categorical imperative of the High Committee for Corporate Governance
In its December 2025 report, the High Committee for Corporate Governance alerts:
“In this context of transformation that is both very rapid and multifaceted, Boards must promote responsible AI governance that is structured and aligned with social and environmental strategy and goals.”
This message is unambiguous: administrators need to take up the AI topic. Not as a technical skill, but as a strategic, ethical and systemic issue. Governing AI means anticipating the value created and the risks, preserving the human being, and aligning uses with the company's raison d'être. It's knowing how to ask the right questions.
Tips: Take a Kantien jump to regain control!
To meet these challenges, Armand Jiptner and Olivier Jochem, experts in board governance, have designed a high-level training course that we co-lead.
A training in 4 parts, where the 4 questions of Emmanuel Kant's philosophy come to our aid. The very architecture of human reason:
1. Collectively understanding AI in the Council (what can I know?)
2. Integrating AI into the Council's strategy and control missions (what should I do?)
3. Propose an “enhanced” governance of the Council in terms of AI (what can I hope for?)
4. Preserve the human factor in AI decisions, in line with the Loi Pacte and the AI Act (what is humans?)
What you get out of it:
· A global approach: from the functioning of AI systems to the challenges of governance and strategy
· Customization to your organization: AI strategic vision, governance, risks and opportunities
· Active sequences: practical cases, committee simulations, strategic questioning exercises
AI is transforming the game. It is time for the Councils to tackle it with lucidity, method... and ambition
