The weeks go by and are not alike in the tech sector. Between artificial intelligence migrating directly into our pockets, European regulations starting to bite, and smartphones changing their philosophy, several topics deserve our attention. Here are the notable tech news and innovations of this week, broken down for clarity.
On-device AI on smartphones: why your next applications will work without a connection
Have you noticed that some functions of your phone, like automatic photo editing or voice transcription, are getting faster, even without a network? It’s no coincidence.
You may also like : The latest must-see celebrity news and scoops
Apple, Qualcomm, and MediaTek are now integrating dedicated artificial intelligence chips directly into their mobile processors. We’re talking about NPU, or Neural Processing Unit: a specialized component that executes AI tasks locally, on the device, without sending your data to a remote server.
In practical terms, this results in embedded generative services: automatic call summaries, intelligent photo editing, voice assistants that respond offline. These functions, long reserved for cloud applications, will be available on models released in 2025 and 2026.
Read also : Fashion Trends: Discover the Must-Have Styles to Embrace This Season
To keep track of the evolution of these topics over the weeks, the news from the Intronaut site regularly covers advancements in artificial intelligence and hardware innovation.
Why this choice of “on-device”? Two main reasons. The first is latency: processing an image or text locally takes a fraction of a second, compared to several seconds via the cloud. The second is privacy. Your data stays on your device and does not pass through a third-party server. For a professional call summary or real-time translation, the difference is tangible.
European AI Act: what the regulation changes for the services you use
Europe is not just watching AI deploy. The European regulation on artificial intelligence, called the AI Act, was formally adopted in 2024. Its obligations will be implemented gradually between 2025 and 2026.
What does this mean in daily life? Systems considered “high-risk” (automated recruitment, credit scoring, biometric video surveillance) must now comply with strict risk management rules. Large language models, which power chatbots and text generation tools, are also affected.
Transparency on training data
One of the most concrete obligations concerns enhanced transparency on training data. Foundation model providers will need to document what data was used to build their AI. Users will also benefit from information rights: knowing if content was generated by a machine, for example.
For tech companies operating in France and Europe, this means a need for adaptation. Small businesses integrating AI into their products are not exempt.
PC Copilot+ and NPU: AI is also transforming laptops
The same “on-device” logic applies to computers. Microsoft is pushing its “Copilot+” PCs equipped with dedicated NPUs, capable of running AI models locally.
The uses far exceed the integrated chatbot in the search bar. Here’s what these machines offer today:
- Contextual local search: find a file, an email, or a web page by describing its content in natural language, without knowing the exact name of the document.
- Real-time subtitling of any video, even offline, with automatic translation into multiple languages.
- Image editing and generation directly within system applications, without third-party software or cloud subscriptions.
The common point among these functions: they run without an internet connection and without sending data to a server. Local AI is becoming both a selling point and a technical choice.
A change of philosophy for manufacturers
Until now, AI computing power resided in data centers. Chip manufacturers (Qualcomm, Intel, AMD) are now integrating neural processing units into their consumer architectures. This transition redistributes the cards: a mid-range laptop in 2026 will have capabilities that would have required a dedicated server three years ago.
For users, this translates to more responsive machines for creative and office tasks. For software publishers, it adds an extra development constraint: optimizing their applications for these NPUs, in addition to traditional processors and graphics cards.
Tech innovation in France: signals to watch this week
Beyond AI, several movements deserve attention in the French ecosystem. The cybersecurity sector remains under pressure, with increasingly sophisticated phishing campaigns targeting SMEs. Protection solutions are evolving accordingly, also integrating layers of analysis through artificial intelligence.
On the mobility front, car manufacturers are accelerating remote software updates for their electric vehicles. The connected car is becoming a fully-fledged tech terminal, with its own cybersecurity and personal data management challenges.
On the pricing front, electronic components are stabilizing after the shortages of previous years. Prices for memory and processors are normalizing, which should gradually reflect on the prices of finished products.
Next week will bring its share of new developments, particularly with anticipated announcements regarding smartphones and connected objects. The common thread remains the same: AI is leaving the cloud to settle in everyday devices, and European regulation is beginning to frame this transformation. Two dynamics that will shape the tech market in the coming months.