X has updated its developer agreement to add a new restriction on the use of posts on the platform for AI training. The updated policy, previously spotted by TechCrunch, states that developers cannot use content from X or its APIs to “fine-tune or train a base or edge model.”
The policy change could force X to enter into AI training agreements with third-party companies, similar to the deal Reddit has with Google. On Wednesday, Reddit, which has a similar policy of blocking AI crawlers, sued Anthropic, claiming that the company’s AI crawlers had accessed the site more than 100,000 times since July 2024. Elon Musk’s AI company xAI acquired X for $33 billion on paper in March.
Although the agreement with X’s developers prohibits companies from training AI on their content, its privacy policy still allows third-party “employees” to train AI models on the site’s data unless users opt out. X also uses user data to train its Grok AI model.
