Meta, the social media giant, recently obtained permission to use the data of its European users to train its artificial intelligence models, while complying with the General Data Protection Regulation (GDPR).
This new measure, which will come into effect on May 27, concerns content shared on Facebook and Instagram, but spares WhatsApp. If you are a user of these platforms, here’s what you need to know and how you can act to protect your data.
Meta AI: What data is affected?
Meta plans to use a wide range of content to refine its AI algorithms. This includes:
- Your posts on Facebook and Instagram, including photos and their captions.
- The comments you leave on these posts.
- Interactions with the Meta AI chatbot via Messenger, such as questions or requests sent.
This data, often personal, is valuable to Meta because it helps improve the performance and relevance of its artificial intelligence tools.
A right to object guaranteed by the EU
In accordance with European regulations, Meta offers its users the option to object to the use of their data for its AI training. The company has started informing users by email, providing a clear procedure to exercise this right.
This approach is inspired by the one already implemented in the United States, but it is advisable to check or renew your objection, even if you have already done so elsewhere.
How to object to Meta using your data?
To prevent Meta from using your data, follow these simple steps:
- Go to the dedicated page set up by Meta.
- Verify that the email address displayed matches the one associated with your Meta account (it is usually pre-filled).
- Submit your request. To strengthen your objection, you can include a link to the CNIL (National Commission for Information Technology and Liberties) website, although this is not mandatory.
- You will receive an email confirming your request within a few minutes.
You don’t need to wait for Meta’s information email to act. You can access the page now and assert your right.
Read also: How can Yiaho offer you ChatGPT for free? Explanation
Why does Meta collect this data for its AI?
For Meta, user data is an essential resource for developing more powerful AI models. It allows for a better understanding of behaviors, preferences, and trends, making tools like recommendations or chatbots more accurate.
Although European regulations require Meta to offer an opt-out option, this collection remains strategic for the company, and the fact that some users object may complicate its ambitions.
Read also: How to recognize AI-generated text?
Act now and be vigilant
By objecting, you prevent Meta from using your personal data to train its AI. However, this does not affect your access to Meta’s AI tools, such as its chatbot. Be careful though: if your friends or contacts do not make the same choice, your interactions with their posts (comments, reactions) could still be used, as they are linked to their accounts.
What about other language models?
The use of personal data by AIs like ChatGPT, Grok, and Gemini raises legitimate concerns.
- ChatGPT, developed by OpenAI, retains all prompts and conversations, which can be used to train its models, unless the user activates “temporary chat” mode or explicitly requests the deletion of their data via OpenAI’s privacy portal.
- On our Yiaho platform, we do not store any personal data. As you know, you don’t even need to register to use our AIs!
- Gemini, from Google, collects up to 22 types of data (such as conversations, location, and browsing history) and retains them by default for 18 months, although this can be adjusted to 3 or 36 months in the settings.
- Grok, created by xAI, is more discreet with only 7 types of data collected, but its use of data from the X platform can lead to biases, and its responses often require verification.
To minimize risks, it is advisable to avoid sharing sensitive information and to regularly check the privacy policies of these tools.
To learn more about your rights, consult the CNIL website or stay informed via our blog for other news on AI and data protection.
