X/Twitter will allow the use of data to train third-party AIs

In a significant change to its privacy policy and terms of service, X (formerly known as Twitter) announced that will allow the use of user data to train business models artificial intelligence third-party (AI). This change, which will come into effect from November 15, 2024, is already generating controversy in the form of the use of user-generated content on social media platforms.

New policies and implications for users

X’s updated privacy policy includes a crucial clause in the “Information Sharing” section. According to the new paragraph, the platform may share or disclose account information with third parties, if users have chosen to release this content. This encompasses the use of information for “independent purposes,” explicitly including training AI models, whether generative or not.

It’s worth noting that while X introduced a setting in July 2024 allowing users to opt-in or opt-out of their data being used to train the platform’s own Grok chatbot, this new change significantly expands the scope to include external companies.

Additionally, X’s terms of service have been modified to include a royalty-free license that gives the company broad freedom to use content posted on the platform. This license allows X “use, copy, reproduce, process, adapt, modify, post, transmit, display, upload, download and distribute” user content.

The new version of the terms specifies that this license includes the right to use the materials to “provide, promote, and improve services,” with explicit mention of training machine learning and AI models. This is a significant change from the previous version of the terms, which did not mention artificial intelligence.

Implications and concerns

This change raises important questions about privacy and intellectual property. Users who disagree with these new policies will have to reconsider their use of the platform, as acceptance of these terms is necessary to continue using X.

The practice of using user posts to train AI models is not unique to X. Other social networks, such as Meta and LinkedIn, also take similar approaches. However, transparency and clarity in communicating these practices have been questioned.

In Brazil, this issue has already generated controversy. The National Data Protection Authority (ANPD) took preventive measures against Meta to prevent the training of AI models with user posts, resulting in the temporary suspension of some generative AI capabilities in the country.

The ANPD also requested clarification from X when the network changed the settings for Grok training. So far, there has been no new information about the municipality’s position in relation to these recent changes in X’s policy.