Loading player
Last September Meta, the company that owns the social networks Instagram and Facebook (as well as the messaging app Whatsapp), made public the first features based on generative artificial intelligence, i.e. on those software which, trained on gigantic masses of data, are capable of generating texts, videos and images that respond to a specific request. These are features that, in one way or another, almost all large technology companies have begun to introduce or at least test in the last two years.
However, to work well, and continue to work better, these artificial intelligence systems require that the mass of data on which they train is progressively expanded. In this Meta is very advantaged by the fact that the platforms it owns contain billions of photographs and posts published by users over twenty years, in dozens of different languages. And in fact last month Meta began using those posts, and especially those images, to train its artificial intelligence system, Meta AI.
When you sign up to Facebook or Instagram you accept the terms of use, which say that Meta has full freedom to modify, copy or create derivative works from all images and videos published on the platform. Contents exchanged via private message and those found within private profiles are excluded, i.e. accessible only to those who ask for access to the owner of the profile in question. According to journalist Mack Degeurin, however, the ability to draw on the content of “hundreds of millions of loyal users who regularly upload data-rich photos and videos to its platforms for free could give Meta a big advantage over the competition.”
The decision was communicated to users with a notification on both Facebook and Instagram, but many people did not pay much attention to it, also because the habit of ignoring company communications on the sites they frequent is quite widespread. Initially, therefore, much of the criticism came from artists and content creators who publish their works on Instagram and Facebook and who are very careful about the risk that the company uses their creations to train Meta AI. The fact that generative artificial intelligences like Midjourney and Dall-e can reproduce an artist’s style quite faithfully, and therefore have a significant impact on the revenues he obtains from his work, is in fact a much-felt and debated topic since these technologies were made public.
In recent days, some posts and videos have begun to circulate widely on Instagram reminding users that it is possible to ask Meta to exclude their posts from the Meta AI training process.
One of the posts on Instagram that explain how to exercise the right of opposition (@selmatically)
In reality you can only do this if you live in the UK or an EU country, thanks to various privacy and data management laws that have been introduced in recent years (such as the GDPR or the more recent Digital Services Act ). Users living in the rest of the world will be able to fill out a separate form: the company said it will review requests but could still choose to reject them and continue using that content.
Even for European users, however, according to some, the system is “intentionally designed to be annoying and try to minimize the number of users who actually end up objecting”. In the case of Instagram it is necessary to go to the Settings section, then click on Information, then on Privacy Policy. On that page you can find various information relating to Meta AI: those who want to object must click on the link on the words Right to object and fill out the form. Once completed, a page appears where Meta says it will “review objection requests in accordance with relevant data protection laws.”
– Read also: Bots are dehumanizing the internet
