Your phone suggests the end of your sentences. Your bank detects a suspicious transaction before you do. Your doctor consults software to refine a diagnosis. In 2024, these situations are no longer new but part of an acceleration affecting regulation, professional practices, and daily life simultaneously. The digital landscape in 2024 is characterized less by the emergence of new technologies than by their entry into binding legal frameworks and large-scale adoption.
AI Act and NIS2: two European regulations redefining the digital framework
You may have noticed that every announcement about artificial intelligence now comes with a legal mention or warning? This new reflex stems from a specific text: the AI Act, which came into effect in August 2024. This European regulation classifies AI systems by risk level and imposes graduated obligations on companies that develop or use them.
A point often misunderstood: the AI Act does not apply as a whole. Its implementation is gradual. Some prohibitions (such as social scoring systems) take effect quickly, while other obligations, particularly for general-purpose AI models, follow a staggered timeline over several years. Presenting this regulation as an immediately applicable constraint to all uses would therefore be inaccurate.
In parallel, the NIS2 directive has set a transposition deadline of October 17, 2024. It expands cybersecurity requirements to a much larger number of sectors and organizations. In France, this transposition is through the so-called “Resilience” bill, which illustrates the gap between the adoption of a European text and its translation into national law.
To keep track of these regulatory and technological developments over time, Cyber sPass articles regularly cover these topics with a practical angle. NIS2 makes cybersecurity a governance issue, not just a tool issue. The directive includes obligations for risk management and incident reporting, with an increased role for executives in oversight.
Generative artificial intelligence: what has changed for businesses
Generative AI has moved from a stage of curiosity to being integrated into business processes. In 2024, use cases have clarified in three main directions.
- Assisted writing in marketing, legal, and communication services, where tools generate drafts that teams refine. The time savings are real, but human proofreading remains necessary to avoid factual errors.
- Internal data analysis: companies use language models to query their document bases in natural language instead of formulating technical queries. A salesperson can ask a question about a customer’s history and receive a usable summary.
- Augmented software development: code assistants (like Copilot) speed up program writing. Developers spend less time on repetitive tasks and more on architecture and business logic.
Generative AI does not replace a job; it changes the distribution of work time. Raw production tasks decrease, while verification, supervision, and strategy tasks increase. This redistribution forces companies to rethink their job descriptions and training paths.
The trap of dependency on proprietary models
A rarely discussed angle: market concentration around a few model providers (OpenAI, Google, Anthropic, Meta) creates technical dependency. A company that builds its processes on a proprietary model is exposed to changes in pricing, terms of use, or performance without notice.
Open-source initiatives (Mistral in France, Meta’s LLaMA) offer an alternative but require more advanced technical skills for deployment and maintenance. The choice between proprietary and open-source models has become a strategic decision, not just a technical parameter.
Cybersecurity and regulatory overlap: what organizations need to anticipate
The new European obligations do not exist in isolation. The AI Act, NIS2, and the DORA regulation (dedicated to the financial sector) overlap on several points: management of risks related to third-party suppliers, incident reporting, documentation of critical systems.
For a financial sector company using AI tools for fraud detection, three regulatory frameworks apply simultaneously. This overlap creates a need for legal mapping that many organizations have not yet realized.
Digital governance: the role of executives is changing
NIS2 introduces explicit responsibility for executives regarding cybersecurity measures. It is no longer a topic that can be entirely delegated to a technical manager. Executives must understand digital risks just as they do financial risks.
This evolution pushes organizations to integrate cybersecurity into their management committees and strategic decision-making processes. Training for executives on these topics is multiplying, a sign that the market has identified this gap.
Sustainable technologies and digital sobriety: beyond rhetoric
The energy consumption of digital technologies remains a contentious issue. Data centers necessary for training AI models consume considerable amounts of energy and water. In 2024, this reality has prompted several players to rethink their infrastructures.
Two approaches coexist. The first focuses on hardware optimization: more efficient processors, liquid cooling, locating data centers in areas with decarbonized energy. The second questions the relevance of certain uses. Do we need to generate an image by AI for every social media post?
Digital sobriety is not limited to an individual gesture. It involves choices in infrastructure, software design, and business models. Companies that integrate this dimension from the design of their digital services gain a regulatory and reputational advantage.
The year 2024 marks a turning point where digital innovations collide with structuring regulatory constraints. The AI Act, NIS2, DORA: these texts transform how technologies are developed, deployed, and supervised in Europe. The next step is not to adopt more technologies but to better govern those already in place.
