On Monday 3 June, the European Data Protection Supervisor (EDPS) published its guidelines on the use of generative AI by European institutions.
The EDPS stresses that, given the rapid development of the artificial intelligence sector, these guidelines do not cover all the different problems that may arise, but offer as broad a view as possible.
The supervisor encourages cautious use of generative AI, particularly with regard to the use of the data collected and the type of model used (see EUROPE 13415/12).
It also reiterates its distrust of ‘scraping’ techniques, the indiscriminate export of personal data, and believes that institutions should, if they don’t develop their own generative AI models, study the options on the market to choose those that comply with the various European regulations.
“It is EUI’s responsibility to appropriately manage the risks connected to the use of generative AI systems. Data protection risks must be identified and addressed throughout the entire life cycle of the generative AI system”, the EDPS explains in detail in its guidelines.
The document is a “first step” in supporting the institutions, says the controller, who is publishing it in his capacity as supervisory authority.
The EDPS is also responsible for supervising the European institutions under the AI Regulation (see EUROPE 13311/17).
To see the orientations, go to https://aeur.eu/f/cit (Original version in French by Isalia Stieffatre)