Full Breakdown
Meta Removes Controversial AI-Training Language from Its Commercial Terms
By Drooid · · How we work
Background and Legal Context
Legal analysts at Nieman Lab described the provision as “remarkable,” suggesting it could allow Meta to scrape any publisher’s site—including news outlets—when the publisher holds a Meta business account, and to use that material to train its artificial-intelligence models. The wording raised concerns that the clause might override a site’s robots.txt instructions, a point highlighted by scholars at Neiman Lab and by James Grimmelmann, a professor of digital and information law at Cornell University.
Official Statements from Meta
In response to the backlash, Meta’s vice president of communications, Andy Stone, and company spokesperson Daniel Roberts issued a joint clarification. They emphasized that the updated language applied only to Meta’s advertising and business products and that the company had not altered its web-crawling practices. Roberts added that content owners could continue to use robots.txt to express their preferences and that the revision did not grant Meta any new rights to scrape news sites.
Criticism from Legal Scholars and Industry Observers
Critics argued that the original clause would have effectively granted Meta a blanket license to harvest and repurpose publisher content for free, undermining existing copyright protections. Their concerns centered on the impact on news organizations that rely on licensing revenue and on broader questions about the ethics of using scraped content to train commercial AI systems.
Implications for Content Owners and AI-Training Practices
The removal of the contentious language narrows Meta’s contractual claims over scraped content, reaffirming that standard web-crawling rules—including robots.txt signals—remain in force. For publishers, the change reduces the risk of involuntary inclusion of their articles in Meta’s AI training pipelines, though the company’s broader data-collection practices continue to be governed by separate policies. Observers note that the episode highlights ongoing tensions between large tech platforms and content creators over the use of publicly available data to develop increasingly sophisticated AI products.