for the CNIL, the Clearview AI algorithm is a “very intrusive data processing”

The algorithm created by this American start-up makes it possible to identify a person by taking a photo of them. The Cnil is demanding a fine of 5.2 million euros from this company for not having paid a fine imposed last October.

Article written by

Published

Update

Reading time : 2 min.

A man taking a photo with his smartphone, in July 2022 (illustration image).  (KARL-JOSEF HILDENBRAND / MAXPPP)

The principle is simple: you take a photo of someone and the application’s algorithm identifies them in a few seconds thanks to all the photos of them that exist on the internet. Clearview AI is based on a colossal database of more than 30 billion public images, the largest in the world to date, according to the New York company.

This database is made up of photos posted by users on social networks such as Facebook, Instagram, Twitter and Youtube. This system is at cœur arm to do between the Cnil and Clearview. “It’s a very intrusive data processingdenounces Karin Kiefer, director of the protection of rights and sanctions at the Cnil. It is considered that this processing is unlawful since the persons have not consented to the processing of their data and, moreover, cannot exercise their right to access the data and obtain their erasure.

“Neither customer nor company in France”

The Cnil is demanding a fine of 5.2 million euros from the American start-up specializing in facial recognition for not having paid a fine of 20 million euros imposed last October. The French regulator blames Clearview “to suck up photographs from many websites, including social networks” and asks him not to proceed, without a legal basis, to the collection and processing of data from people in France, and to delete the data of these people after having responded to the access requests sent to him. The company refuses to comply with this compliance.

>> Olympic Games 2024: the Cnil calls on parliamentarians not to introduce facial recognition into the law

Clearview, founded in 2017, does not intend to change its practices or pay the fine requested by the CNIL. “Cooperation between France and the United States is slow and complex, explains lawyer Alexandre Lazarègue, specialist in digital law. Under these conditions, obviously, it makes decisions very difficult to execute in the United States.”

In a statement sent to franceinfo, the founder of Clearview AI said “sorry for the erroneous interpretation by some of the nature of the activities of his company which has neither a client nor a company in France”. Among its customers, Clearview AI has counted in the United States local police, immigration services, customs or private companies.

For the CNIL, the Clearview AI algorithm is a “very intrusive data processing” – report by Sandrine Etoa

listen


source site-29