A financial director receives a fine for AI non-compliance, while the implicated model was deployed by an external provider. Since August 2026, this scenario is no longer theoretical. The transparency obligations of Article 50 of the AI Act apply to companies using AI systems, not just those designing them. For decision-makers, the trends and innovations to follow directly engage the operational responsibility of the company.
AI Act Compliance: What Decision-Makers Must Check Right Now
There is a lot of talk about generative artificial intelligence in executive committees. There is less discussion about the regulatory framework governing its daily use. Since February 2, 2025, organizations deploying AI must ensure that the relevant personnel are proficient in the tools they use. Since August 2, 2026, transparency obligations extend to chatbots and certain AI-generated content.
Specifically, this means that a customer service department using a chatbot must inform the user that they are interacting with a machine. A marketing department generating visuals through AI must indicate this. AI compliance is no longer solely the responsibility of the legal department, but of every business function.
For leaders of SMEs and mid-sized enterprises, the risk is to assume that compliance falls on the model provider. The AI Act distributes responsibilities throughout the chain. One can discover the latest updates on Infos Décideur to keep track of these evolving obligations as the implementing texts become clearer.
- Identify all AI uses within the company, including tools integrated into third-party software (CRM, HR, accounting).
- Ensure that each use visible to a customer or employee complies with the transparency obligation applicable since August 2026.
- Document the training of teams in mastering the AI systems they handle, as required by the text since February 2025.
Cybersecurity for SMEs: The Gap Between Discourse and Actual Coverage
Cybersecurity has been at the top of strategic concerns for companies for several years. Field feedback shows a less reassuring reality. The Hiscox 2026 report reveals that French small and medium-sized enterprises are reducing their cyber insurance coverage. The proportion of companies without a dedicated policy is increasing.
This gap poses a concrete problem. An SME that suffers a ransomware attack without adequate coverage absorbs the shock on its cash flow. The amounts at stake often exceed the self-financing capacity of a structure with fewer than 50 employees.
Why SMEs Are Scaling Back on Cyber Insurance
Several factors are at play. Premiums are rising, exclusions are multiplying in contracts, and many leaders mistakenly believe that their size protects them. Reducing cyber coverage at a time when attacks are diversifying is akin to removing the fire extinguisher because one has never had a fire.
For a decision-maker, the question is not whether cybersecurity is a priority (everyone answers yes), but whether the allocated budget truly covers risk transfer. An audit of the existing cyber policy, cross-referenced with a mapping of sensitive data, takes a few days and can prevent a liquidity crisis.
Data and Skills: Two Projects That Condition Everything Else
One cannot talk about digital transformation without addressing data quality. A Business Intelligence tool or a generative AI model is only as good as the data that feeds it. In practice, many companies accumulate data without clear governance: duplicates between CRM and ERP, outdated customer files, HR data scattered across multiple spreadsheets.
Without data governance, every new technological tool amplifies the existing disorder. Before investing in an artificial intelligence or cloud project, the priority is to clean and structure the data assets. This work is often invisible, rarely valued in executive committees, yet it is crucial.
Hybrid Skills: What Recruitment Reveals
The job market is evolving towards profiles that combine business expertise and mastery of digital tools. A management controller who knows how to query a database, a logistics manager capable of configuring a real-time dashboard: these hybrid skills are becoming a recruitment criterion, not just a bonus.
Feedback varies on this point across sectors, but the underlying trend is clear. Investing in the digital skills development of existing teams is cheaper than a failed external recruitment. Continuous training programs are justified here by a direct economic calculation, not by an HR stance.
Security and Technological Innovation: Arbitrating Without Dispersing
Cloud, IoT, no-code tools, process automation: the list of available technologies is growing each year. For a decision-maker, the trap is to pile up projects without prioritizing. Companies are often observed launching a cloud migration, an AI project, and an ERP overhaul simultaneously, with the same teams and budget.
The result is predictable: none of the three projects is completed on time. It is better to carry out one transformation project to completion than to have three stalled in parallel.
- Identify the project that generates the quickest operational return (reduction of manual tasks, time savings on a critical process).
- Allocate a dedicated team with a clear mandate, rather than spreading the workload across already mobilized employees.
- Measure the results after six months before launching the next project.
Security must accompany every step. A cloud migration without a data encryption policy or a business continuity plan exposes the company as much as it modernizes it. Technological innovations do not replace the fundamentals of risk management.
The 2026 trends for corporate decision-makers converge on the same observation: the available technology far exceeds the absorption capacity of most organizations. The real competitive advantage does not come from the adopted tool, but from the rigor with which the groundwork is prepared, whether it concerns regulatory compliance, cyber coverage, or data quality.
