Your phone suggests how to finish your sentences. Your watch detects an abnormal heart rate before you feel anything. Your neighbor’s smart fridge orders milk on its own. These small changes, accumulating over just a few years, are reshaping how we work, consume and protect our data. Three areas of technology are now attracting particular attention: artificial intelligence regulation, mandatory sustainability standards for devices, and the infrastructure that powers it all.
EU AI Act: what it means for everyday apps
Have you noticed that some apps now display a warning when content is generated by artificial intelligence? This isn’t a marketing choice. Since February 2, 2025, the EU AI Act has prohibited certain practices considered high-risk, such as social scoring and large-scale behavioral manipulation.
Since August 2, 2025, the rules have also applied to general-purpose AI models. In practical terms, a software provider that integrates a language model into its service must document the training data used and assess systemic risks. Transparency is becoming a legal requirement, not a selling point.
For users, this means clearer technical information about AI tools, more visible opt-out options, and clear accountability when an algorithm makes a decision that affects them (hiring, credit, diagnosis). Companies that publish tech trend analysis on the Kali News website are keeping a close eye on these regulatory developments, as they will shape the real-world adoption of these tools.
Repairable smartphones and eco-design: sustainable tech becomes law
Why replace your phone every two years if it still works? For a long time, the answer came down to two words: planned obsolescence. Glued-in batteries, updates that slow down the device, and parts that are impossible to find.
Since June 20, 2025, European eco-design rules have changed the game. Manufacturers of smartphones and certain tablets must ensure spare parts remain available for several years. Security software updates must also continue well beyond the standard warranty period.
What appears on the energy label
Devices covered by the rules now display a repairability rating on their energy label. Before buying, you can compare two models on this specific criterion, just as you would compare the energy consumption of refrigerators.
The expected impact goes beyond simple repairs. Using a device for one extra year reduces its environmental footprint far more than careful recycling. Manufacturers must rethink their business models, which have traditionally relied on frequent replacement. Some see this as a constraint; others see it as a selling point for premium ranges designed to last.
Data center energy consumption and generative AI
Every request sent to an AI assistant uses energy. Not just the electricity powering your screen, but also the energy consumed by the data center processing the request, cooling the servers and storing the results. The expansion of generative AI has led to a sharp increase in power demand.
Ten years ago, a 30-megawatt data center was considered a major facility. Today, projects exceed one gigawatt, according to industry players such as ABB. The question is no longer whether AI is useful, but whether the infrastructure can keep up.
Nuclear power, solar energy and liquid cooling: current approaches
Data center operators are exploring several ways to meet this demand:
- Connecting directly to low-carbon energy sources (nuclear power and large-scale solar) to power facilities without relying on the conventional grid during peak hours.
- Using liquid cooling for servers instead of traditional air conditioning, reducing the energy needed to keep server rooms at the right temperature.
- Optimizing AI models themselves, since a lighter model requires fewer computing cycles to deliver an equivalent response.
These infrastructure choices are not abstract. They determine where a data center can be built (access to water, proximity to a power source), how much a cloud subscription costs and, ultimately, the price you pay for an online service.
Agentic AI and automation: when software acts without human approval
Today’s AI assistants answer questions. The next generation, known as agentic AI, plans and carries out tasks from start to finish. You give it a goal (arrange a business trip, screen job applications, monitor inventory), and the agent handles each step without intervention.
Gartner identifies this trend as one of the most transformative for organizations. The difference between an agent and a simple chatbot lies in its decision-making autonomy: it compares options, places orders and follows up with suppliers.
The risk of automation without safeguards
Would you trust a program to handle your purchases and negotiate on its own? This is not a theoretical question. Early implementations show that without clear limits, an agent may optimize one criterion at the expense of another (the lowest cost, for example, but with an unreliable supplier).
Defining the boundaries of an AI agent’s actions is just as strategic as choosing the tool itself. This ties directly into the EU AI Act mentioned above: an agent making automated decisions in a high-risk area (healthcare, finance, hiring) will have to meet requirements for traceability and human oversight.
- The scope of the agent’s actions must be defined in advance, with human approval thresholds for sensitive decisions.
- The agent’s activity logs must remain accessible for internal or regulatory audits.
- If an error occurs, responsibility lies with the company deploying the agent, not the provider of the underlying AI model.
Agentic AI is no laboratory curiosity. It is making its way into management tools, customer service platforms and accounting software. The difference between a successful deployment and a costly incident often comes down to how carefully the initial scope is defined, not the power of the model.
The tech trends of this period are about more than new gadgets. They are reshaping responsibilities: manufacturers’ responsibility for sustainability, software providers’ responsibility for algorithmic transparency, and companies’ responsibility for the energy they consume. Our digital lives are changing as much through regulation as through innovation itself.
