The year 2024 has reshuffled the cards in the high-tech sector, not so much due to the arrival of a flagship product but rather because of a simultaneous regulatory and industrial shift. Generative artificial intelligence has spread into consumer devices, smartphone manufacturers have revised their offerings around chips dedicated to embedded AI, and the European Union has established a binding legal framework with the AI Act. These parallel movements are shaping a technological landscape where product innovation can no longer be dissociated from regulatory compliance.
European AI Act: the regulatory framework redefining tech in Europe
Regulation (EU) 2024/1689 on artificial intelligence, which came into force on August 1, 2024, now structures the design and marketing of any AI system deployed in the Union. Its implementation timeline extends until 2028, but the first obligations have already had concrete effects on products sold in Europe.
Since February 2025, practices classified as “unacceptable risk” are prohibited: social scoring, subliminal manipulation, certain forms of emotional recognition. These prohibitions directly affect voice assistants, connected speakers, and wellness applications that exploited behavioral data without safeguards.
Field feedback varies on how Asian and American manufacturers are integrating these constraints into their product cycles. Some have anticipated by limiting features on European models, while others are adapting on the fly.
The obligations specific to general-purpose AI models (GPAI), applicable from August 2025, add an additional layer: technical documentation, adversarial testing, and enhanced respect for copyright. For brands that integrate generative assistants into their smartphones or televisions, each software update becomes a compliance exercise.
Starting in August 2026, transparency obligations will come into force for any system interacting with the public. Users must be informed that they are interacting with AI, and synthetic content must be clearly identifiable. This aspect concerns both chatbots integrated into e-commerce sites and automatic photo filters on smartphones. The resources available on the DigiGeek.fr high tech site allow tracking of these developments product by product.
Embedded artificial intelligence in smartphones: beyond marketing talk
The most visible trend of 2024, on the consumer side, remains the integration of artificial intelligence directly into smartphone chips. Apple has designed its A18 Pro chip to accommodate the features of Apple Intelligence, while Qualcomm and MediaTek have multiplied the NPUs (Neural Processing Units) in their SoCs intended for Android devices.
The promise is enticing: local processing of requests, automatic photo editing, real-time transcription, contextual suggestions. In practice, the majority of these functions still depend on a cloud connection for complex tasks. Purely local processing often limits itself to photo scene detection or battery optimization.
The available data do not allow for a conclusion on a real gain in autonomy for the average user. Manufacturers’ benchmarks measure the raw power of the NPU, not its daily utility. However, the arrival of the AI Act requires that these embedded processes comply with transparency rules, which could hinder certain behavioral profiling functions integrated into software layers.
What this means for device renewal
Manufacturers are using embedded AI as a premium upgrade argument. A 2023 smartphone without a dedicated NPU will be presented as obsolete, even if its performance remains sufficient for most uses. Embedded AI accelerates the renewal cycle without proportional benefits for the user.
Transparent screens and connected objects: showroom innovations or real products
CES 2024 in Las Vegas showcased transparent screens, notably from LG and Samsung. These panels allow information to be displayed while still being see-through, with applications envisioned in home automation, retail, and automotive.
The technology has existed for several years in prototype form. What has changed in 2024 is the announcement of initial commercializations in specific niches:
- Augmented shop windows, where the transparent screen replaces static signage while allowing visibility of the products displayed behind
- Semi-transparent office partitions integrating video conferencing and data display, tested in a few corporate headquarters
- Extended heads-up display windshields in the automotive sector, where the screen covers a much larger area than current HUDs
The limitations remain clear: insufficient brightness in daylight, high production costs, and lack of interface standards. No consumer model is expected for several years.
Generative technologies and software development: a skills transfer in progress
Generative AI is no longer limited to text writing or image generation. In 2024, its adoption in software development has crossed a visible threshold. Code assistants (GitHub Copilot, Amazon CodeWhisperer, tools integrated into JetBrains IDEs) have moved from experimental status to daily tools in many tech companies.
This transfer raises unresolved intellectual property questions. Generative models trained on open-source code sometimes produce excerpts nearly identical to existing repositories. The AI Act imposes technical documentation on the training data of GPAI models, but concrete control mechanisms remain to be defined.
For companies integrating these tools, the measured productivity gains on repetitive tasks (unit testing, documentation, refactoring) coexist with an increased risk of technical debt. Code generated without a deep understanding of architecture can create dependencies that are difficult to maintain.
Impact on digital professions
Desired profiles are evolving: the ability to formulate precise prompts, audit generated code, and understand the limitations of models is becoming a recruitment criterion. Development training is gradually incorporating these skills, without consensus on their exact place in curricula.
The high-tech landscape of 2024 is characterized by this tension between technical acceleration and regulatory oversight. The European AI Act sets a timeline that manufacturers can no longer ignore, while embedded AI and generative AI are reconfiguring uses without their real benefits always being demonstrated. The coming months, with the implementation of obligations on GPAI models in August 2025, will provide a first concrete measure of the impact of this framework on the products available in Europe.
