The social media platform X may be looking to license its content to third-party AI companies

Reports have emerged that the social media platform X may be looking to license its content for AI training as an alternative source of income. X published an update stating that it made some changes to its Terms and Privacy Policy.

In a rather vague language, the update specifies that, among several other changes, X has "added language" to its Privacy Policy which specifies how X may use information users share on the platform to train AI models. Further, X also made changes to the language specifying how long it stores information, and included information on how users in the European Union can challenge some of X's decisions under the Digital Services Act.

When one goes to the Privacy Policy, it turns out that one of the changes regarding the use of information concerns third-party usage. According to the updated policy, agreeing to share your information with X now includes agreeing to having X share your information with "third-party collaborators" who can use the information for their own purposes, including training AI models. The wording is not the clearest, but for now, it seems that users can either opt in to everything, or, if they don't want their information shared with third parties, opt-out of sharing their data entirely.

As for the changes that describe X's data retention policies, it formerly stated that profile information and content were stored for the duration of the account, while personally identifiable information was stored for up to 18 months. The updated version simply states that it keeps different information for different periods of time, depending on the type of information and its purpose, such as providing its products and services, complying with legal requirements, and ensuring safety and security.

Finally, X's Terms of Service have been updated to include a "Liquidated Damages". Tellingly, this section specifies that anyone caught "requesting, viewing, or accessing more than 1,000,000 posts in any 24-hour period" will become liable to X for liquidated damages, to the tune of $15,000 USD per 1,000,000 posts. Furthermore, those agreeing to X's Terms must recognize the charge as reasonable, not consider it a penalty, and accept it as not limiting the company of pursuing other legal actions, including equitable relief and statutory damages.

As the demands of fresh training data intensify, most sources are tightening their grip on their content, fighting off scrapers, and negotiating licensing deals. X is no exception, as it trains its Grok assistant using X user data and has famously eliminated the then Twitter API free tier to stop "bot scammers & opinion manipulators". By licensing its content, X would potentially open a much needed source of income given its troubles getting advertisers back into its platform and its struggles replacing advertisement income with paid subscriptions earnings.