Meta Ai is the IA chatbot which collects the most user data. Can the latter still regain control over their private information?
Applications used daily by billions of people, Facebook, Instagram, Messenger and Whatsapp have one thing in common: they belong to Meta, a web giant led by its main founder, Mark Zuckerberg. If Meta has never been exemplary in the processing of user data, the recent integration of a conversational artificial intelligence (AI) in some of its applications should only worsen the loss of user control a little more.
Meta AI: An assistant at the service of users …
Meta Ai first appeared in the United States in 2023, then in France in early 2025. At the end of January, a META press release said that this technology already had more than 700 million monthly active users. Meta Ai is based on the Llama language model (Large Language Model Meta Ai) developed by Meta. This improved chatbot is positioned as a conversational assistant. He advises users, provides them with information and offers recommendations. Users can also use it to generate content (text summary, image, etc.). Meta AI wants to differentiate itself from its competitors via the personalization of the content offered thanks to the analysis of user behavior and interactions. According to a recent study, this tool is the IA chatbot which collects the most data on users, with 32 types of data out of 35, more than double the average (13).
… Or Meta’s right arm?
WhatsApp messaging, which had 3 billion users worldwide in the first quarter of 2025, welcomed Meta AI a few weeks ago. By opening the application, users discovered a blue circle at the bottom right of their screen. When they click on it, a conversation can start with this assistant who is not unanimous. Indeed, Meta plans to use the videos, photos or comments of certain WhatsApp users in order to train its AI models. This project had been announced in 2024 but then suspended by the Irish data protection authority (DPC) which called into question the transparency of data processing. Europeans are not the only ones to question these practices. In July 2024, Brazil had asked Meta not to use the data of Brazilian citizens to cause its models.
WhatsApp users are concerned about respecting their privacy and also fear that their personal data, communicated during discussions with Meta AI, be used for commercial, advertising or publicly exposed purposes in the case of cyber attack. Meta Ai considers the lack of reaction of users as an implicit consent. However, the General Data Protection Regulation (GDPR) requires obtaining explicit consent for any processing of personal data.
Regain control
How to allow users to easily and long -term use their data control? The easiest way today is to suggest that they do not interact with Meta AI – which only activates once it has been requested. If there is a form to oppose this functionality, it is difficult to access and is “dark patterns”, an interface which has been voluntarily designed to deceive or manipulate a user.
To be sure to exchange in peace with their loved ones or their colleagues, users can also opt for another messaging as a signal (70 million users) or Telegram (950 million users). These two applications have recorded a significant increase in downloads in recent years, in particular thanks to features which provide maximum security of personal data (end -to -end encryption, no data collection, no advertising or tracker, etc.). Proof that for some users, the confidentiality of their data remains essential.
