Meta has quietly updated its commercial terms of service, removing a clause that explicitly stated the company could use content from external domains and URLs to train and improve its products. The removed language had granted Meta broad permission to scrape data from websites, including news organizations, to develop its AI features. This update comes after significant backlash from industry observers and legal experts who raised concerns about the implications of such data collection practices.

While Meta has not issued a formal statement explaining the specific reasoning behind the removal, the change is widely viewed as a response to growing pressure regarding copyright and data usage in the AI sector. Critics, such as legal scholar James Grimmelmann and reporting from outlets like Nieman Lab, had previously highlighted the language as a potential overreach that could undermine the intellectual property rights of news publishers. By striking this clause, Meta has effectively narrowed the explicit scope of its data collection terms, though the company has not clarified whether this indicates a change in its actual scraping operations or merely a shift in its legal documentation strategy. The move highlights the ongoing tension between AI developers and content creators over the use of web-scraped data for training large language models.