© Lemon squeezer
While we are still waiting for the arrival of Meta AI in Europe, the firm formerly known as Facebook is beginning, according to our colleagues at Android Authority, to train its model on the # 8217;huge database of the social network. Nothing in itself very surprising… except that certain users targeted by this collection begin to receive a notification. And that a tweet on 𝕏 describes the incredible complexity of the procedure to follow to refuse this massive collection of data.
This tortured journey gives us the effect of a surprise, because it's not really as if Meta hadn't slipped up, in this matter, in the past . We still remember the Cambridge Analytica scandal and its consequences very well. Here, the aim is obviously to reduce as much as possible the number of users refusing this collection. Which is probably, incidentally, contrary to certain provisions of texts in force in Europe such as the DMA, the DSA or the recently passed IA Act.
Steps to refuse Meta AI training on your personal data
However, the famous notifications are starting to appear on certain accounts United States – and that we know the steps you should not miss to refuse the training of Meta AI based on the data on your account. Note that these steps are translated from English. The exact wording may differ slightly on accounts based in France:
- From your computer, locate a Meta notification saying “We are planning new AI features for you. Learn how we use your information”, then click on it (otherwise it is not possible to go to the second step).
- You will then see a policy update page with only a button Close which YOU MUST NOT CLICK.
- Instead, locate a link that says something like “right to opt out” or “to’object” (at least in English version) and click on it.
- This takes you to a form in which it is important to fill in all the boxes – including, very importantly, the description of the refusal of this use of your personal data.
- Validate the form; unfortunately this is not the end of the journey: you must retrieve a single-use code from your emails and enter it on the page before validating.
- From there, an immediate response explains that Meta will examine your request… which can take a variable length of time.
- If your refusal is validated (a little strange to formulate in this way but let's move on), you will receive another email confirming to you that Meta will not use your personal data for its AI; it's best to keep a copy of this email somewhere just in case.
A priori, the biggest risk that currently arises around the training of these AIs on content from the internet is the question of copyright . Especially when it comes to image generation. When it comes to training on content from a social network, however, we enter terra incognita in terms of risks. Even if 𝕏, with his AI Grok, is probably already into it. However, we are not talking about the same proportions: the Facebook platform alone has nearly 3 billion users.
(2.5 billion users). #8217;users), YouTube (2 billion users) or even Instagram (2.5 billion monthly active users). Another universe when we know that Elon Musk's social network has “only” 225 million active users, according to Linda Yaccarino (2023 figures).
- Meta starts sending notifications to certain users.
- This is a policy change around AI, suggesting that the firm begins training its AI on data generated by users on its platforms.
- It is possible to refuse this data collection for the first users to receive the notification – at the cost of a series of quite complex and uncertain steps.
📍 To not miss any news from Presse-citron, follow us on Google News and WhatsApp.
< /blockquote>
[ ]