X Alters Policy to Ban Use of Its Content for AI Model Training

The social network X has revised its agreement terms with developers to prohibit third parties from utilizing the platform’s content for training large language models.

Notably, the section «Reverse Engineering and other Restrictions» has been updated to include a clause that explicitly forbids the use of the «X API and X Content for fine-tuning and training any foundational or advanced models.»

These modifications followed xAI’s acquisition of X in March for $33 billion, a firm owned by Elon Musk. This acquisition allows xAI to potentially limit competitors’ access to the social network’s data.

It is likely that the platform will enter into agreements for model training, similar to the actions taken by Google and Reddit. The latter has a policy against AI scraping, which led it to sue Anthropic for violations.

In 2023, X altered its privacy policy to allow the use of publicly available data on its site for training AI models. In the fall of last year, the company made further amendments to permit third parties to train their models.

Despite these adjustments, X’s privacy policy still maintains that external “partners” may train AI models on platform data unless users explicitly opt out. Additionally, X leverages user data to train its own AI model, Grok.