Since August 2, 2026, the European regulation on artificial intelligence (AI Act) has reached a milestone. The enforcement powers of the AI Office and national authorities are activated. For leaders and managers, this means a concrete shift from monitoring to operational compliance.
AI Act and Recruitment: What the Enforcement Phase Changes for HR Teams
Are you using CV screening software or application scoring tools? This type of tool falls into the category of high-risk AI systems according to the AI Act. The use of automated pre-selection, matching, and profile scoring is now at the center of European compliance concerns.
The timeline deserves careful reading. The heavy obligations related to high-risk AI systems have been postponed: some systems in Annex III will only be fully affected by December 2, 2027, and others, integrated into regulated products, by August 2, 2028. This delay provides time, but it does not exempt from acting now.
The obligations for transparency are already applicable. A candidate interacting with a recruitment chatbot must be informed. AI-generated content must be labeled. Companies that publish HR analyses on the site Le Blog des Décideurs regularly remind us: identifying internal AI tools is the first step towards compliance.
Decision Mapping: Distinguishing What AI Can Decide Alone
Éric Hazan, former director at McKinsey and co-author of “Should We Still Decide?”, proposes a simple framework. Every company should create its own decision map, classifying each decision according to two criteria: the relevance of delegation to AI and its acceptability.
Relevance is high when three conditions converge: data is abundant, the metric to optimize is clear, and the human dimension is low. A pricing algorithm on an e-commerce site meets these conditions. An annual performance review does not.
The Trap of Poorly Defined Co-Decision
Many teams today operate in co-decision with AI without formalizing it. A manager asks a generative tool to draft a performance summary, then validates it in a few minutes. This process poses a specific problem: the anchoring bias pushes to validate the machine’s proposal rather than questioning it.
Éric Hazan puts it this way: the main risk with AI is stopping to think. The manager must maintain what he calls his “decision-making muscles,” meaning the ability to formulate an independent judgment before consulting the generated response.
In practice, this requires a simple protocol:
- Draft your own analysis or recommendation before launching the AI request, even in the form of quick notes
- Systematically compare your conclusion with that of the tool, identifying discrepancies
- Document cases where the human decision diverged from the algorithmic suggestion, to feed into a collective feedback loop
Supplier Failures and Payment Delays: Two Economic Signals to Monitor
The French economic news of mid-2026 reveals a tension that financial statements do not always show. An Ivalua report indicates that supplier failure poses an increasing risk to supply chains. Payment delays between companies exacerbate this fragility.
Why does this issue concern management as much as finance? Because a failing supplier does not only translate into an accounting line. It generates team reorganizations, emergency production arbitrations, and pressure on operational managers who have not been prepared to handle this type of crisis.
How an Operational Manager Can Anticipate Supplier Risk
Integrating a financial health indicator of key suppliers into quarterly team reviews changes the game. This is not a natural reflex for a production manager or site director. The subject is traditionally handled by procurement or the finance department.
Three concrete actions can reduce exposure:
- Request a simplified dashboard from the procurement department on the average payment times of strategic suppliers
- Identify, for each critical supplier, a backup plan that can be activated within two weeks
- Include a regular point on supply continuity in management meetings, just like performance tracking
Hybrid Management and AI Skills: Adapting Practices Without Reinventing Everything
The HR trends of 2026 confirm an underlying movement. Hybrid management (combining in-person and remote work) remains the norm in most French companies. The upskilling in AI adds to this reality without replacing it.
The trap would be to treat these two subjects separately. A manager who trains his team in the use of AI tools in person but does not provide any support for remote employees creates a skills asymmetry. Access to AI training must follow the same logic as access to information: available regardless of the workplace.
The question of regulating AI usage in the company intersects with daily management. Who validates the use of a new generative tool in a team? Who checks that customer data is not injected into an external model? These decisions do not solely fall under the IT department. The frontline manager becomes the first filter for AI compliance within their scope.
The economic and managerial news of this fall 2026 converges towards the same observation: tools are evolving faster than governance frameworks. Decision-makers who map their decisions, monitor their supply chain, and train their teams in AI in a structured way gain a competitive edge. Those who wait for the regulatory deadlines of 2027 and 2028 risk having to catch up with an organizational backlog that is more costly than compliance itself.
