X updates privacy policy to allow third-party AI training on user data
On Wednesday, social media platform X (formerly Twitter) revised its privacy policy to allow third-party collaborators to use user data to train artificial intelligence models. This change, effective November 15, 2024, means that unless users opt out, their posts could be utilized by external AI companies, a move that mirrors the strategies of platforms like Reddit and various media organizations in licensing data to AI providers.
Policy Changes and Data Sharing
The updated privacy policy introduces a new section on "Third-party collaborators" under its "Sharing Information" clause which states that X may share user information with third parties for purposes beyond those previously outlined in the company’s policy.
Specifically, this could include using the data to train AI models, whether generative or otherwise. The language suggests that AI companies could have access to vast amounts of user content, sparking concerns over privacy and data security.
The policy notes that users can prevent this data-sharing by opting out, though details on how to do so remain vague. Currently, under the “Privacy and safety” settings in X, users can toggle off data-sharing with xAI’s Grok AI chatbot and other business partners, though those partners are described as collaborators helping X “run and improve its products.” Hopefully, the opt-out process for third-party AI providers will become clearer once the updated policy takes effect in November.
Broader Implications
This update comes after controversy surrounding X owner Elon Musk's use of the platform's data to train his xAI’s Grok chatbot. The move prompted an investigation by the European Union’s lead privacy regulator.
While X previously made no mention of sharing data with outside AI firms, the new policy signals the company's intention to monetize its user-generated content by licensing it to third parties.
X has been looking for additional revenue streams following a wave of advertiser withdrawals, boycotts, and the company’s struggle to launch its subscription services.
Liquidated Damages for Content Scraping
Additionally, X has introduced a "Liquidated Damages" clause aimed at curbing unauthorized scraping of its content. The new rule stipulates that organizations scraping more than 1 million posts within a 24-hour period will be liable for damages of $15,000 USD per 1 million posts.