X changes policy to restrict AI training on its data

X’s new policy represents a major reversal from its 2023 stance, when it first allowed public posts to be used for AI training and later expanded third-party access for this purpose, marking a significant shift in data usage permissions within the same year

Social media platform X, formerly known as Twitter, has updated its developer agreement to prohibit the use of its content and API data for training large language models (LLMs).

In the revised terms, published on June 4, the company explicitly states that developers are not allowed to use X's API or content to fine-tune or train any foundational or frontier AI models. The new clause falls under the "Reverse Engineering and Other Restrictions" section of the agreement.

X joins data protection trend

The move follows X’s acquisition by xAI, Elon Musk’s artificial intelligence venture, earlier this year. Analysts suggest the change is aimed at preventing rival AI firms from freely leveraging X’s vast user-generated content without a formal licensing agreement.

This policy marks a significant shift from X’s earlier stance. In 2023, the platform had revised its privacy policy to permit the use of public posts for training AI. By late that year, third-party access for training purposes had been further enabled.

X now joins a growing list of tech companies, including Reddit and The Browser Company, that are tightening control over data access in response to rising concerns around AI training and content ownership.