Meta removes disputed AI-training language after publisher scraping backlash
The Verge reports that a broad content-training clause has disappeared, while Meta says its crawler practices and publishers' robots.txt preferences have not changed.
A controversial clause disappears
Meta has removed language that legal experts warned could let it use publishers’ websites to train AI through their business-account terms, The Verge reported on October 17. Meta says the revision clarifies existing ads and business-product crawling terms and does not change its web-crawler practices.
The disputed provision said Meta could use content from associated domains, web pages and URLs to develop, improve, research and train its products and services. Nieman Lab’s October 15 report, based on three legal experts’ review, warned that the language could be read as permission to scrape news websites whose publishers held Meta business accounts. The upcoming terms were scheduled to take effect on October 30.
That was a legal interpretation of the wording, not a court ruling or evidence that Meta had begun a new scraping operation. Meta disputed that reading before the latest revision. The Verge subsequently reported that the broad content-use sentence no longer appeared on the commercial terms page.
Meta says crawler preferences still apply
In a follow-up statement quoted by The Verge, spokesperson Daniel Roberts said the consolidated crawling terms applied only to ads and business products. He said content owners could continue signaling preferences through robots.txt and that the updates did not give Meta new rights overriding those preferences. TechCrunch also reported the removal and the company’s clarification.
For publishers, the concrete development is a change to language whose scope had raised concerns about training rights. Meta’s response is its stated position, not an independent audit of how its crawlers behave. The removal also does not settle broader questions about copyright, AI training or whether earlier wording would have been enforceable.
AITrending read the English reports and quoted company response. A direct fetch of Meta’s terms URL returned only a Facebook page shell, so the clause’s removal is attributed to The Verge’s page comparison rather than presented as our own complete legal-document comparison.