X updated its privacy policy to let the company use collected data to train AI models; Elon Musk says X may use “just public data, not DMs or anything private”
X's recently updated privacy policy informed its users it would now collect biometric data as well as users' job …
Context & Ripple Effects
This policy change follows X's addition of biometric, jobs and education data collection to its privacy terms and Musk's stated interest in using public tweets for xAI. It turns the platform's user-data rules into a direct input to its AI strategy.
The boundary is consequential because X subsequently paired restrictions on outside scraping and crawling with a later policy allowing third-party AI collaborators to train on user data. Together, those moves point to X treating platform data as a controlled asset rather than an openly harvestable web resource.
First-order effects
- X gains an explicit policy basis to use collected platform data for its own model training, while users face a broader stated use for information they provide or generate.
- Musk's public-data limitation distinguishes public posts from DMs and other private material, but leaves users to assess how the revised policy applies to other collected data categories.
Second-order effects
- AI developers seeking X-like conversational data face a more uneven access landscape: X can train internally while its terms constrain unapproved collection by others.
- The change raises the value of first-party platform data and increases pressure on social platforms to specify whether user content is reserved for internal models, shared with partners, or excluded from training.
Third-order effects
- If this pattern holds, training-data access will increasingly be governed by platform terms and negotiated permissions rather than by the practical availability of public web content.
- The durable fault line becomes the permission boundary for third-party AI training: distinctions among public posts, private communications and other account data are likely to shape user trust and scrutiny of AI-data practices.
The trend: Social platforms are converting user-generated data from an implicitly open web resource into a governed input for proprietary and partner AI models.