Meta removes model training clause from its commercial terms after scraping concerns rise

Meta withdrew a legal clause it recently added to its commercial terms of use after it sparked widespread concerns among publishers and legal experts about turning business accounts into an open permission to scrape website content and use it to train artificial-intelligence models. The amendment reported by specialised media contained explicit wording giving the company the right to exploit any content taken from domains, web pages or links associated with a business account for development, research, training and to maintain the safety of its products and services.

The clause alarmed press organisations and content creators, as a report by Nima Lab citing legal experts described it as an exceptional measure that could allow the company to crawl any publisher’s sites that have a business account on its platforms and pull their material to feed language models.Legal dispute reveals data source sensitivity in AI industryNotably, the company overcame a major lawsuit last year over intellectual-property infringement, but it still faces multiple lawsuits with other companies over the legality of training models on web content without explicit permission.

Amid rising objections, a report by The Verge said Meta removed the controversial clause, stating in an official press release that it responded to feedback and rewrote the terms to clarify that the unified crawling rules apply only to advertising and business products, emphasizing that the move does not change how its web crawlers operate. Although a company spokesperson initially insisted the update was not intended to alter indexing or data-collection mechanisms, the swift withdrawal reflects a growing awareness of institutions’ rejection of using advertising gateways to pass hidden training rights.

This rapid retreat highlights the shift of commercial terms of use into a silent testing ground for acquiring training data, after most sites blocked public web crawlers via robots-txt files.The current equation makes advertising contract clauses a tool no less sensitive than technical software.Major tech platforms seek to secure new data streams through shared-service agreements, while publishers cling to defending their digital archives and exclusive content from unpaid consumption.

This trajectory directly impacts the business environment in the Gulf, Egypt and the Levant, where hundreds of media organisations, e-commerce platforms and start-ups rely on Meta business managers to run marketing campaigns and technical integration with social networks. The latest development forces executive leaders and marketing teams in the region to break the common habit of automatically accepting terms-of-service updates and to subject these changes to rigorous legal and technical review.Data protection no longer limited to securing internal databasesIt now also requires ensuring that linking website domains to external platform accounts does not create a contractual loophole allowing the use of Arabic content to train its algorithms without separate, direct licensing agreements.