Meta is updating the terms and conditions of its privacy policy to allow it to scrape data to train artificial intelligence models.
As written, the updated terms and conditions will require companies to let Meta use automated software to “access, index, cache, analyze, collect, retrieve, or crawl the domains, webpages, or URLs” associated with the company in question. This includes the ability to create “test credentials,” which are sometimes used to bypass paywalled content.
The updated policy applies only to Meta’s business-related products and accounts. It won’t affect normal users, whose public posts already fall under Meta’s standard privacy policy, which allows the company to scrape those posts to train AI models.
When first posted, the updated terms explicitly said Meta could use the scraped data to “train” its products, strongly implying the data would be fed into AI models. The word “train” was removed from the terms sometime between Oct. 15 and 20, so it’s unclear whether, or how, that slight language change affects what Meta may do with the data.
In October 2026, reporting from Harvard University’s Nieman Journalism Lab revealed that Meta, the technology giant that owns and operates Facebook, Instagram, and WhatsApp, among others, planned to alter its commercial terms of service. The change would allow the company to freely collect data from the internet using bots and crawlers—a practice commonly known as “scraping”—to train artificial intelligence models. The documentation indicated that these updated terms were scheduled to take effect on October 30.
An investigation confirmed that the Nieman Lab’s reporting was largely accurate. The proposed modifications did appear to incorporate a clause permitting Meta to harvest data and online information without paying licensing fees, including content from news organizations. Consequently, the claim was rated as mostly true.
However, following the publication of the Nieman Lab report, Meta altered the wording within the proposed update. This modification raised questions about how it might influence the company’s capability to utilize that information for AI training purposes.
In electronic and verbal communications, Meta spokesperson Daniel Roberts asserted that the revised terms and conditions did not alter the company’s data collection or usage practices. Instead, he described the new paragraph as a routine consolidation of pre-existing policies into a single location.
Roberts also contested the notion that the updated language would facilitate the unauthorized extraction of content from news websites. “As always, content owners can use robots.txt to signal their preferences,” he noted, referencing the configuration file that instructs web crawlers on site accessibility. “This is not about scraping news sites, and these updates don’t give us new rights that supersede those preferences.”
Key details regarding the proposed adjustments include the following:
Meta’s Business Model
To fully grasp the significance of these proposed updates, it helps to examine Meta’s primary revenue stream, which relies heavily on advertising. In essence, external businesses invest substantial budgets to run promotional campaigns across Meta’s social media platforms.
Because advertising constitutes the vast majority of its income, Meta is heavily motivated to streamline and optimize the ad-buying process for enterprises. This is accomplished primarily through two main strategies.
First, advertisers naturally seek the widest possible audience, making user volume a critical metric. However, user growth is finite once a platform approaches market saturation. Third-party data indicates that Meta’s platforms attract roughly 3 billion active monthly users, suggesting the company has largely maximized the physical size of its consumer network.
Second, advertisers demand robust data to precisely target specific demographics and measure the return on investment for their campaigns. Meta’s extensive and intricate privacy policies allow the company to gather and retain a vast array of user data.
These combined practices position Meta as a comprehensive advertising hub. Rather than selling user data outright to third parties, Meta retains the information internally, leveraging its proprietary software and algorithms to determine which ads reach specific audiences. Additionally, Meta provides tools that enable businesses to experiment with and refine their marketing strategies using algorithmic support.
Due to its immense scale, Meta’s ecosystem is virtually indispensable for small and medium-sized digital enterprises—including publishers who utilize the platform to promote newsletters and membership programs through paid ad space. Naturally, participating in Meta’s business ecosystem requires agreement with specific terms, conditions, and privacy frameworks.
The previous terms and conditions (archived) for Meta’s commercial products were designed primarily to limit the company’s legal liability. Under those guidelines, participating businesses granted Meta an intellectual property license for advertising purposes, committed to complying with relevant U.S. laws regarding data sharing, and released Meta from liabilities stemming from third-party actions.
The Changes Highlighted by Nieman Lab
The updated terms (archived) introduced several novel sections that, if accepted, could significantly broaden Meta’s operational authority.
The most critical element appeared in Section 1, captured in the version reviewed by Nieman Lab on October 15:
You authorize us (or any entity acting on our behalf) to use automated software (such as web scrapers, bots, robots, spiders, crawlers, and user agents) and create test credentials to access, index, cache, analyze, collect, retrieve, or crawl the domains, web pages, or URLs (and any content available through them) that are associated with your products or services, or that are made available through your or users’ use of Meta Company Products and services, and to periodically pull updates from such domains, web pages, or URLs in order to provide, develop, improve, research, train, and maintain advertising and business products… and maintain the integrity and security of Meta Company Products.
In straightforward terms, this clause initially appeared to grant Meta authorization to extract data from the websites of any business utilizing its commercial tools, including material typically restricted by paywalls.
Legal professionals interviewed by Nieman Lab expressed shared concerns regarding the scope of the provision.
“This is a remarkable provision to bury in terms of service, given that this would give Meta access to material that would likely cost many millions of dollars in a licensing market,” stated intellectual property attorney Matt Topic.
“For news publishers, this clearly represents permission to Meta to train AI on any content from their sites and possibly to deploy some of that content, even in competition with the publishers themselves,” noted Cornell University digital law professor James Grimmelmann.
Despite these interpretations, Meta maintained that the revisions merely consolidated existing policies without expanding data access rights. Roberts reiterated that website operators retained the option to block data collection via their robots.txt configurations.
Reports from Ars Technica noted that while older terms permitted scraping only under narrow conditions, the updated language was substantially broader, potentially allowing data collected for one purpose to be repurposed in entirely different ways.
It is worth noting that various publishers have filed lawsuits against major artificial intelligence developers, including Meta, alleging that copyrighted material was systematically harvested without compensation to train AI models.
The Subsequent Modification
The interpretation that Meta was securing rights to scrape corporate data specifically for AI training hinged heavily on the inclusion of the word “train” within the commercial terms. The relevant phrase originally read: “…in order to provide, develop, improve, research, train, and maintain advertising and business products.”
However, when the terms were archived on October 20, the word “train” had been quietly removed, a change also reported by The Verge on October 17.
When questioned about the removal of the word and its implications for data usage, Meta provided a brief statement:
“We removed ‘train’ because there was some confusion. As we previously said, the consolidation of terms were for business and advertising products, not scraping news websites.”
Follow-up inquiries regarding the practical effect of removing the term were not directly addressed by the company spokesperson, leaving the ultimate impact of the language adjustment unclear for website operators evaluating the new terms.