Launch of the Centre pour la Sécurité de l'IA (CeSIA)

We are delighted to announce the launch of an initiative that we have been working on for several months: the Centre pour la Sécurité de l'IA (CeSIA).

‍

CeSia is an evolution of EffiSciences' AI division and specializes in technical research and engineering.

Why this initiative? Artificial intelligence is experiencing rapid development, opening up immense possibilities but also involving major security risks. Problems such as jailbreaks, hallucinations, lack of transparency, cybersecurity issues, or even the loss of control of AI systems, are bound to take on worrying proportions given the speed at which they gain in capacity and autonomy.

‍

Faced with a global preparation that seems insufficient, it is crucial to promote the engagement of a variety of actors. This is precisely the mission that CeSIA has set itself: to support key AI players to prepare them for future challenges, by contributing to the establishment of a rich and collaborative French AI ecosystem.

‍

The core business of CeSIA will be to provide technical insight into AI development trends, in order to identify current risks and issues and to anticipate future challenges. Our activities will focus on three main areas:

• Research: We will develop open source tools to monitor the operation of AI models and detect failure modes, including those still unknown to developers.
• Information for the general public: We are committed to raising collective awareness about the uses and challenges of AI, so that everyone can understand and get involved in these crucial issues.
• Training future generations of researchers and developers, with an emphasis on evaluating and securing AI systems.

‍

CeSIA aims to play a key role in building a future where AI will be both responsible and safe, for the benefit of humanity.

‍

The CeSIA team

‍