How to prevent Meta from using your photos and texts posted on Instagram and Facebook to train its AI

Meta is updating its privacy policy with information about how it is using the content published by users on its social networks to train artificial intelligence models. This update will come into effect on June 26 and, along with it, Meta has enabled a form so that those who wish to do so can exercise your right to object and prevent, although not completely, the content that they have generated over the years, on platforms such as Facebook and Instagram, be phagocytosed by AI.

As the company explains, in relation to the information with which it trains its artificial intelligence models, 'we use publicly available online information and licensed information. We also use the information you shared in Meta products and services. For example, posts or photos with their descriptions. We don't use the content of your private messages with friends and family to train our AIs.' User interactions with Meta chatbots, available on a limited basis in some countries, are also part of the training data.

The new privacy policy includes Instagram, Facebook, Messenger and Threadsamong other Meta products, according to the list of affected platforms indicated by the company, but not to WhatsAppwhose last update in its privacy policy is from last February, without any reference to artificial intelligence.

Meta also explains that in that public information with which they train their models there may be personal data, such as name and contact information, but which are not linked to any Meta account. He also acknowledges that There are cases where it will continue to use user data to train AI even if the right to object is recognized or even without having a Meta account. For example, if another user shares an image in which you appear or if he provides information about you in posts he shares.

If you want to prevent content generated by you on Meta networks from being fodder for AI, you must access this link. You have to select your country of residence, provide an email address and answer two questions. The first, mandatory, 'Explain how this data processing affects you'; the second, optional, 'Please provide additional information that may be helpful to us in reviewing your objection.'. Meta assures that it will review objection requests 'in accordance with relevant data protection laws' and that if it accepts 'will apply from now on'.

The generative AI tools that Meta plans to bring to all its products in different formats create text and images after being trained with gigantic volumes of text, image and audio data. “By studying this information, you can learn, for example, the relationship and associations between different types of content. In this way, models can create new content when a person gives them instructions or asks them a question,' explains Meta.

So now you know, if you don't want your photographs, texts, ideas or creations published over the years on Facebook and Instagram to be part of the material with which the Meta AI is trained, now you can exercise that right.