NIXSolutions: Facebook’s New AI Training Policy

The social network Facebook has announced changes to its privacy policy regarding the use of user photos and messages to train artificial intelligence. The changes will come into force on June 26, 2024.

As PCWorld reports, starting on that date, messages, images, and other public content from millions of Facebook users will by default be available for analysis by Meta’s (Facebook’s parent) neural networks. This is necessary to improve the performance of generative AI tools. At the same time, personal messages and correspondence will not be used – access to them will continue to be limited.


User Concerns and Rights

Experts note that such a step was quite expected. Mark Zuckerberg’s company has been increasing investments in the development of AI technologies for several years, which require huge amounts of different data to train.

The concern is that Facebook is now effectively accessing people’s information without their direct approval. Although the company itself claims that the innovation will benefit not only it, but also its users, part of the social network’s audience has already expressed their dissatisfaction.

In this regard, Meta emphasizes that every user has the “right to object.” This can be done using a special form, which is offered on the help page. In the form, you must indicate your data and the reason for the objection. Meta is then obliged to consider the objection and evaluate it in accordance with applicable data protection legislation. However, even if the objection is upheld, Facebook could still use some of the data to train the AI, such as when other people share a user’s content or mention it in their posts.

The use of publications by social network participants to train AI could have been foreseen by the end of last year, notes NIXSolutions. However, a new agreement, unprecedented in scale, comes into force for the first time. We’ll keep you updated on any further developments related to this policy change.