Meta has delayed enforcement of its revised business terms until November 20, moving back an October 30 start date after publishers objected to language they feared could authorize broad web crawling for AI-related uses. The Meta business terms delay leaves the underlying dispute unresolved: the company says the rules do not expand its scraping rights, while legal experts say the remaining language may still be broad enough to cover AI training.
Daniel Roberts, a Meta spokesperson, told The Verge that the company wanted more time to explain to partners and stakeholders what would change and what would not. Roberts also said Meta would use the extra time to explain its web-crawler practices.
The disputed terms apply to business accounts and would authorize automated software to access and crawl web domains, pages and URLs associated with a business's products or services, according to reporting by The Verge and the Mercury News. Earlier wording reportedly said Meta could use material from those sites to provide, develop, improve, research, train and maintain its products.
Why did Meta delay its business terms?
Publishers raised concerns that the terms could let Meta claim permission to collect material from news sites and use it in AI product development. The News/Media Alliance wrote to Meta chief executive Mark Zuckerberg seeking revisions and warning of harm to publishers, The Verge reported. Publishers and digital advertisers in the European Union and the UK also urged antitrust regulators to suspend the terms, according to the outlet.
Meta removed wording that explicitly referred to training its products, The Verge reported. That did not settle the argument. Cornell Law School professor James Grimmelmann told Nieman Lab that the terms' reference to material associated with a company's products or services could permit scraping of whole sites. Media and intellectual-property lawyer Matt Topic told Nieman Lab that permission to use scraped material to develop an undefined business product could be broad enough to include training.
Meta has rejected that interpretation. Roberts told The Verge the update was not intended for scraping news sites and did not grant rights that override robots.txt preferences. He said content owners can use robots.txt, a file through which site owners can signal crawler preferences.
That assurance remains a point of contention rather than a resolved contractual fact. The Mercury News reported that robots.txt provides voluntary instructions to automated systems. Legal experts cited by The Verge said the terms' language did not clearly rule out the use of crawled material for AI training.
Which version of Meta's terms will apply?
The text itself has produced conflicting reports. PPC Land said that when it checked Meta's UK-English terms preview on October 18, the document still included the earlier language allowing crawled material to be used to train Meta products. Other reports said Meta had removed the training reference. PPC Land said the available material did not explain the discrepancy, which could reflect different language editions, timing or publication status.
Meta's new November 20 date postpones enforcement. It does not, on the reporting available, answer which final language will govern or whether the company will further revise the terms before then.
This story draws on original reporting from The Verge.
