Artificial intelligence: the Cnil creates a division responsible for analyzing risks


The National Commission for Computing and Liberties (Cnil) is taking the plunge into artificial intelligence. ChatGPT, DALL-E, Stable Diffusion and soon Sparrow are among the AI-based models that have been popping up on the web in recent months. Ever more efficient, these technologies fascinate as much as they worry. In an attempt to understand the issues and risks associated with these new tools, the CNIL is creating a division specially dedicated to the issue.

What framework for AI?

The new field of action of the French policeman of personal data will be attached to the Direction of technologies and innovation composed of five people. Lawyers and several engineers will work together to understand the consequences of AI in more detail and try to better regulate it. Documentation for professionals and the general public should be available soon. The Cnil will also study in depth the risks to privacy linked to this technology, in conjunction with the probable entry into force in the coming months of a European regulation. ad hoc.

“The use of this technology therefore concerns all players, whatever their fields of activity or their size. By extension, everyone is necessarily directly or indirectly confronted with it at one time or another. For example, when using a voice assistant, a job application selection process or in the fight against tax evasion”details the commission which wants to establish good practices for the sector.

the deep learning (also) concerned

Alongside their work on artificial intelligence, the experts will also study the impact of deep learning (deep learning) on the personal data of French users. As a reminder, this process used upstream of the development of an AI consists of processing a large amount of data present on the Internet (from data brokers, users or open sources) to feed the learning of the algorithms used. In its statement, the administration discusses the privacy risks associated with this crucial step in the development of an AI. Many public and private actors have already contacted the Cnil on the issue for their own needs.

By studying deep learning and AI in its general context, the CNIL hopes to publish tools over the year to “supporting the creation and use of databases”as well as “practical sheets to answer the most common situations encountered by learning database users”while waiting to interpret the future European regulation.



Source link -98