X is changing how its Privacy Policy describes the use of user data, and the update puts artificial intelligence training at the center of the shift. The social network, formerly Twitter, now says third-party collaborators may be able to use information from X for their own AI models unless users opt out.
The change matters because it extends the AI data question beyond X owner Elon Musk’s xAI and its Grok AI chatbot. According to the updated policy language, outside recipients may use shared information for independent purposes, including artificial intelligence training.
What X changed in its Privacy Policy
On Wednesday, X updated its Privacy Policy with new language in Section 3, titled Sharing Information. The added paragraph focuses on third-party collaborators and explains that user information may be shared or disclosed depending on a person’s settings or their own decision to share data.
The key passage says: “Third-party collaborators. Depending on your settings, or if you decide to share your data, we may share or disclose your information with third parties. If you do not opt out, in some instances the recipients of the information may use it for their own independent purposes in addition to those stated in X’s Privacy Policy, including, for example, to train their artificial intelligence models, whether generative or otherwise.”
That wording signals a broader data-sharing path than the one already associated with Grok. X owner Elon Musk trained xAI’s Grok AI chatbot on X user data, which led to an investigation by the EU’s lead privacy regulator. Before this update, the company had not amended its policy to say that X data might also be used by third parties.
The policy points users to X’s settings page, but it does not identify a specific place inside settings where people can disable this new kind of data-sharing. At the moment described in the source article, the Privacy and safety area lets users turn data-sharing on or off for xAI’s Grok and for other business partners.
The opt-out is central, but still unclear
The new policy language depends heavily on the idea that users can opt out. The practical issue is that the source article says X’s policy does not yet clearly show where that opt-out control lives for third-party AI collaborators.
The existing business partners setting is described as covering companies X may work with to run and improve its products. That description is not the same as saying data can go to other AI providers for their own independent model training.
One possible reason is timing. The updated privacy policy will not become effective until November 15. The source article notes that X could add the opt-out option by then.
For users, the plain-language takeaway is simple: the updated policy says data may be used by third parties for AI training unless the user opts out, but the exact setting path was not specified in the policy language described by the article.
Retention language also changed
X also removed older wording about how long it keeps certain user information. The deleted paragraph said X keeps user profile information and content for the duration of the account, and keeps other personally identifiable data collected through its products and services for a maximum of 18 months.
The replacement language is broader. It says X keeps different types of information for different periods of time, based on how long the company needs to retain it to provide products and services, comply with legal requirements, and handle safety and security reasons.
The updated section gives an example involving usage information. It says content users post, and interactions with other people’s content, will be kept for the duration of the account or until the content is removed.
X also added a note about what can happen after public content is deleted or expires on X. The policy says search engines and other third parties may keep copies of posts for longer, based on their own privacy policies.
Data licensing and scraping sit on opposite sides of the update
The policy update suggests X is exploring data licensing as a new revenue stream for AI companies. The source article compares this direction to moves by Reddit and various media organizations, which have also looked at licensing data to AI companies.
At the same time, X added a new Liquidated Damages section to its updated Terms of Service. That section targets organizations that scrape X content at large scale.
The new terms say an organization will be liable for damages if it requests, views, or accesses more than 1,000,000 posts in any 24-hour period. X says the organization will be charged $15,000 USD per 1,000,000 posts. The covered posts include reply posts, video posts, image posts, and any other posts.
Together, the changes draw a sharper line around X data. The company is opening a policy route for approved third-party collaborators to use data, including for AI training, while also putting a price on large-scale scraping that happens outside its permitted channels.
Why the change matters for X
The move comes as X looks for new ways to generate revenue. The source article notes that advertiser withdrawals and boycotts, along with a subscription feature that has yet to take off, have left the company needing new ways to pay its bills.
That context helps explain why user data is now being framed as something that could support AI-related business relationships. For AI companies, social network posts can be useful because they reflect public conversation, reactions, and interactions. For X users, the update raises a direct privacy question: whether their posts and activity may help train models beyond X’s own products.
X did not respond to a request for comment, according to the source article. That leaves the policy language itself as the clearest available guide to what is changing before the November 15 effective date.