Stories
Meta Removes AI Training Clause After Web Scraping Backlash
- •Meta removed a contentious provision from its commercial terms that sparked concerns about using third-party website content for AI training.
- •Legal experts and publishers criticized the wording, warning it could allow Meta to harvest data while overriding robots.txt access restrictions.
- •Meta maintained the wording did not signal a change in its crawling practices and stated that website owners can still use robots.txt.
- •The incident highlights ongoing tensions between tech firms seeking AI training data and copyright holders concerned about unauthorized use.
Key Questions Answered
Meta removed the provision following scrutiny from legal experts who questioned the broad wording. Critics argued the text could be interpreted as permitting Meta to collect third-party website content for AI training without consent or payment, potentially overriding publishers' restrictions.
Robots.txt is a file used by website operators to communicate instructions to automated crawlers regarding which site parts can be accessed. The disputed commercial terms raised concerns that Meta's language might take precedence over these access instructions, though Meta maintains the revised wording does not supersede robots.txt preferences.
Meta faced a high-profile copyright lawsuit over its use of books to train AI models, which a US federal court ruled in the company's favour in 2025. The court found that the plaintiffs failed to establish that Meta's use of their works was unlawful, though ongoing litigation involving Meta and other AI companies continues.
