The year 2024 has reshuffled the cards in the technology sector on several simultaneous fronts. Between the entry into force of the European regulation on artificial intelligence, the rise of autonomous AI agents, and the increasing tensions over cybersecurity in supply chains, the tech landscape is no longer just a race for performance. The regulatory framework, practical uses in businesses, and technical limitations paint a more nuanced picture than the spectacular announcements from CES in Las Vegas.
AI Act: when European regulation redefines innovation
Regulation (EU) 2024/1689, published in the Official Journal of the European Union on June 12, 2024, constitutes the world’s first binding legal framework dedicated to artificial intelligence. Its logic is based on a classification of AI systems by risk level, with graduated obligations ranging from outright bans (social scoring, subliminal manipulation) to simple transparency requirements for low-risk uses.
This text does not regulate a company as a whole. Each AI system must be evaluated individually based on its use and deployment context. In practice, this requires organizations to inventory their tools, document each use case, and integrate traceability requirements into contracts with their AI suppliers.
Compliance thus becomes an operational technological trend. Technical teams must train employees, implement internal audits, and adapt their development processes. For SMEs, the cost of compliance remains an unknown, and feedback from the field varies on the ability of small structures to absorb these new constraints without dedicated support.
A blind spot remains: civil liability in case of an AI error is not covered by the AI Act. The regulation governs safety and transparency, but the question of who compensates a patient harmed by a misdiagnosis or a candidate excluded by a faulty recruitment algorithm remains open before national courts.
By following the news on the Techronix site, one can gauge how much this regulatory dimension now influences the technological choices of French companies, far beyond just the AI sector.
Autonomous AI agents: beyond text generation
Generative AI dominated discussions in 2023 with the production of text, images, and code. In 2024, the shift is towards agentification. Models no longer simply respond to isolated requests: they chain tasks, interact with other systems, and make intermediate decisions without human validation at every step.
The most advanced application cases concern supply chain management and industrial predictive maintenance. An AI agent can analyze sensor data, trigger an order for spare parts, and schedule a technical intervention, all in a coordinated manner.
The concrete limits of autonomy
This growing autonomy raises questions that the available data cannot yet resolve. The reliability of chain reasoning remains uneven across domains. An effective agent in logistics may produce aberrant results in a medical or legal context, where acceptable margins of error are much lower.
Companies deploying these agents must arbitrate between several constraints:
- The degree of human supervision to maintain, which varies by sector and the risk level defined by the AI Act
- The quality and freshness of training data, which directly affects the relevance of automated decisions
- Integration with existing systems, often designed for sequential processes rather than for agents capable of acting in parallel
Agentification transforms AI from a production tool into a decision-making actor, fundamentally changing how technical teams design their software architectures.
Cybersecurity and digital supply chain: the attack surface expands
The development of remote work and the increasing interconnections between suppliers have significantly broadened the attack surface for companies. In 2024, the cybersecurity of the digital supply chain is emerging as a governance issue, no longer just a technical problem.
AI plays a dual role in this area. On one hand, it enables the detection of abnormal behaviors on networks with a responsiveness that human teams alone cannot achieve. On the other hand, cyberattacks enhanced by generative AI are becoming more sophisticated, particularly in creating large-scale personalized phishing campaigns.
Evaluating supplier risk differently
Traditional cybersecurity approaches relied on protecting the internal perimeter. This logic no longer holds when a company depends on dozens of cloud providers, third-party APIs, and AI models supplied by external vendors. The most advanced organizations are implementing continuous threat exposure management, which continuously maps vulnerability points across the entire value chain.
- One-time security audits are giving way to real-time monitoring of software dependencies
- Contracts with AI suppliers now include traceability and incident notification clauses
- Training non-technical teams to detect AI-related threats is becoming a priority in cybersecurity plans
Robotics and energy: physical infrastructures catch up with software
Software innovations capture media attention, but advances in AI-based robotics and energy infrastructure condition the viability of these technologies in the medium term. The domestic and industrial robots showcased at CES 2024 integrate perception and decision-making capabilities powered by embedded AI models.
In contrast, the energy consumption of AI-related data centers poses a structural problem. The rise of generative models and autonomous agents requires increasing computing capacity, reigniting the debate on nuclear energy as a power source for digital infrastructures.
The technological trends of 2024 should not be read as a list of products to buy. They outline an ecosystem where regulation, security, and energy constraints weigh as heavily as the raw performance of algorithms. The companies that will benefit from these innovations will be those that understand that compliance and robustness now matter as much as speed of adoption.
