Meta Backtracks on Controversial Data-Scraping Language in Business Terms
Meta removes ambiguous contractual language that suggested the company could scrape publisher websites for AI training, following legal pushback.
Last verified:
The Retraction
Meta has removed a clause from its commercial terms of service that sparked immediate concern among legal experts about the company’s authority to harvest publisher content for AI model training. According to TechCrunch AI, the removed provision stated Meta could “use any content from [a business account’s] domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of” Meta products and services.
The language raised red flags because it appeared to grant Meta blanket permission to scrape any website operated by a publisher maintaining a business account on the platform—a practice that could feed training data directly into the company’s AI systems. Legal scholars interviewed by Nieman Lab characterized the clause as a “remarkable provision” that seemed to extend scraping authority well beyond reasonable business necessity.
Meta’s Clarification and Reversal
Meta disputed the interpretation almost immediately. A company spokesperson insisted the updated terms did not signal any operational shift in how Meta crawls the internet. Now, The Verge reports, Meta has formally retracted the language.
In a statement, Meta said: “We’ve heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we’ve clarified that in the terms themselves. It does not change how Meta uses web crawlers.” The company framed the removal as a clarification rather than a reversal, suggesting the provision was always narrower in scope than critics interpreted—confined to advertising and business services, not general web-crawling for model training.
Why This Matters
The episode underscores the legal fragility of AI companies’ data-sourcing justifications. Although Meta defeated a high-profile copyright infringement lawsuit last year, The Verge notes that other cases against Meta and competitors continue through the courts. Ambiguous contractual language that could be interpreted as licensing broad scraping becomes a liability in that environment: it invites litigation and regulatory scrutiny even if the company’s actual practice is narrower.
For publishers and platform operators, the incident also signals that scrutiny of terms-of-service fine print around AI training is now mainstream. Meta’s backtrack, voluntary or not, suggests that explicit, defensible language around data use is becoming a competitive necessity—not just a legal nicety. Companies that fail to draft clear boundaries between ad-targeting and model-training data pipelines risk a repeat of Meta’s reputational friction.
Frequently Asked Questions
What was the controversial language Meta removed?
Meta's updated terms stated the company could use content from business accounts' domains and web pages to train and maintain AI products. Legal experts flagged this as potentially permitting broad web scraping.
Did Meta's actual practices change?
No. Meta stated the removed language only reflected existing practices for ads and business products, and did not represent a change in how the company crawls the web.
Why does this matter legally?
Ambiguous contractual language around AI training data sourcing has become a flashpoint in copyright litigation against AI companies, including Meta, which continues to face ongoing lawsuits.