The CNIL is monitoring the deployment of systems using AI algorithms, in particular by assessing their potential impact on the use of personal data. Specific precautions must therefore be taken to train AI models and also to assert the rights of individuals. Matthieu Camus will discuss the CNIL's recent recommendations in this area.

Matthieu Camus - Privacy Impact

Large-scale language models (LLMs) are emerging as powerful levers for automation and productivity in many organisations. Their ability to adapt to a variety of use cases is as appealing as it is challenging. Indeed, integrating them into operational systems raises a number of issues in terms of personal data protection.
How can we ensure compliance with the RGPD when implementing these technologies? What privacy risks need to be identified, assessed and managed at each stage of the LLM lifecycle?
In this conference, Jérôme de Mercey will analyse the main challenges linked to the protection of privacy in the context of LLMs, and will suggest ways of reconciling innovation, efficiency and compliance.

Jéôme De Mercey - DASTRA

This conference recalls the obligations linked to international data transfers and the need in certain cases to draw up transfer impact analyses (TIAs), with a practical focus on the methodology applicable to such transfers.

Maël Fablet - EY

See Conference

In France, the CNIL wants to extend its scope to include AI. The DPO in the company/administration would then naturally take responsibility for the compliance of such data processing.

What's more, DPOs are already asking themselves the question of how to supervise personal data when it is used in AI systems. This supervision is just as necessary when the data is used during interaction with the device as when it is reintegrated into the learning process to improve the system.

Translated with DeepL.com (free version)