Artificial Intelligence (AI) has continued to establish itself as an essential revolution in our societies, disrupting both our daily interactions and industrial processes. The emergence of models such as GPT-3, accessible to the general public in 2022, has precipitated this change. Generative AI is becoming a reference tool in areas as varied as content creation, data management and virtual assistance. But this technological advance has also revealed areas of uncertainty in terms of discrimination, data security and transparency in particular, which require strict supervision. Europe has taken these issues into consideration by adopting the AI Act, a unique piece of legislation at the global level.
I. The European challenge of trustworthy AI
The challenge for Europe is twofold: protect citizens against violations of their fundamental rights while promoting technological innovation. The adoption of the AI Act, published in the Official Journal on July 12, 2024, embodies this ambition. The AI Act marks a major step forward towards the creation of trustworthy AI, based on three pillars: ethics, legality, and robustness.
II. Risk-based legislation: towards efficient and responsible AI
The AI Act adopts a risk-based approach, itself divided into four levels.
- AI Systems (SIA) at minimum risk – such as virtual assistants or e-commerce recommendations – are only subject to Recommendations.
- The Generative SIA are classified in the low-risk SIA category, with the following only obligation to notify to users that they interact with an AI.
- The SIA considered high risk – which establish individual profiles with consequences on private life (obtaining a loan, a diploma, etc.) – must comply with strict documentation requirements. They also include systems governed by specific European regulations (toys, aviation, etc.).
- Finally theAI Act bans SIAs with unacceptable risks, namely, those who manipulate behavior, exploit people's vulnerability, perform social scoring or use real-time remote biometric identification in public spaces.
For companies like Luminess that develop and deploy AI systems for various industries, This regulation represents both a challenge and an opportunity to strengthen the confidence of its customers.
III. Luminess: Pioneer in AI Act Compliance
Luminess has already taken steps to address this issue. In mapping the SIA and the data used For their developments, the company ensures not only the conformity of the products, but also to be able support its customers as best as possible in this legislative transition.
Each system is analyzed according to its risk level, with a particular focus on security, transparency and data governance. With this approach, Luminess positions itself not only as a responsible supplier, but also as a strategic partner, in a complex regulatory framework.
Conclusion: towards efficient and responsible AI
The AI Act marks a turning point in the history of artificial intelligence. Luminess is committed to being a pioneer in the design of human, ethical and reliable AI., responding to growing expectations for transparency and security. Beyond strict compliance, this challenge reflects a broader ambition: building an AI that serves the collective interest.
By Barbara Bouloux Pariente and Tony Bonnet