Facebook is taking a step back with generative AI. The social network is currently scaling back its plans: there will be no training of its AI using data from its members.
Facebook plays appeasement with artificial intelligence. The social network has made the decision to “ to suspend » for an indefinite period its project consisting of training its AI with the content of its members. In a press release published on June 14, the Irish counterpart of the Cnil confirmed the interruption of this plan.
“ Data Protection Commission welcomes Meta’s decision [la maison mère de Faceook, NDLR] to suspend its plan to train its extended language model using public content shared by adults on Facebook and Instagram in the EU. » Discussions took place between the DPC and Meta.
The Data Protection Commission is Meta’s priority contact at the European level, since it is in Ireland that the social network manages the bulk of its activities on the Old Continent. And it is also in Ireland that NOYB, an association which makes life difficult for internet giants, had filed a complaint against Facebook’s project (as well as 10 other countries).
The Irish CNIL, “ in cooperation with other EU data protection authorities, will continue to engage with Meta on this issue », He added. In addition to Ireland, actions were launched in Germany, Austria, Belgium, Spain, Greece, Italy, Norway, the Netherlands, Poland and France.
Dodging Consent
In the battle for generative AI, Meta is seeking to gain momentum against its rivals, such as Google, Microsoft and now Apple. The social network would have liked to be able to mobilize the enormous reservoir of data available on Facebook, made up of content published by its billions of members – in addition to Facebook, Instagram was also targeted.
Meta’s plan was to train its systems on posts, photos, captions, comments, metadata, and other public content. However, private conversations on Messenger were excluded. For this, Meta had imagined a legal stratagem to avoid prior consent: legitimate interest.
However, the platform offered two forms (one for Facebook, one for Instagram) to object anyway. Meta’s plan regarding generative AI and its members’ data was to come to fruition from June 26, with the arrival of a new privacy policy.
If you liked this article, you will like the following: don’t miss them by subscribing to Numerama on Google News.