Meta revised its commercial terms after publishers and legal experts warned that the original wording could permit widespread scraping of news websites. Critics say the updated policy still leaves important questions unanswered.
Meta has removed language about artificial intelligence training from its upcoming commercial terms following concerns that the policy could permit the company to collect content from news websites without separate licensing agreements. The revised terms, published Friday, October 16, 2026, still authorize automated website crawling but limit the stated purposes to advertising and business products.
The changes followed reporting by Nieman Journalism Lab, which examined how Meta’s proposed terms could affect publishers operating business accounts on Facebook, Instagram and other company platforms. Three legal experts reviewing the original language warned that it could be interpreted as permission to collect news content for AI development.
The revised policy is scheduled to take effect October 30, 2026. Although Meta removed the word “train” and narrowed the description of permitted uses, legal experts say the remaining language could still allow extensive collection of website content.
Meta disputes that interpretation. The company maintains that the revisions clarify existing permissions rather than introduce new data collection practices or override website owners’ preferences.
What Meta Changed in Its Commercial Terms
The controversy centers on a section of Meta’s Commercial Terms describing the permissions businesses grant when using the company’s services.
The original proposed language authorized Meta and companies acting on its behalf to use automated software to access, index, analyze, collect and retrieve material from websites associated with business accounts. It also allowed that material to be used for developing, improving, researching and training Meta products.
That wording raised concerns because it extended beyond material businesses directly uploaded to Facebook or Instagram. It potentially covered content hosted on separate websites, including news articles published outside Meta’s platforms.
After the concerns became public, Meta revised the provision. The new version removes the explicit reference to training and identifies advertising and business products as the permitted purposes for the collected information.
However, the authorization to use automated software remains. The terms still describe tools such as web scrapers, bots and crawlers that can collect information from associated websites and periodically retrieve updates.
The provision also permits Meta or its representatives to create test credentials when accessing websites. Legal experts have questioned how broadly that authorization might apply, particularly when information is located behind registration systems.
The updated text does not specifically limit automated collection to online stores or product catalogs. Instead, it refers to websites and pages associated with a business’s products or services.
That distinction is important because news organizations, independent publishers and other businesses routinely connect their websites to Meta accounts for marketing, audience engagement and advertising.
Legal Experts Say Concerns Remain
Although the removal of the word “train” addresses one concern, specialists in technology and intellectual property law argue that the revised language remains unusually broad.
Cornell Law School professor James Grimmelmann told Nieman Journalism Lab that the additional wording did not resolve his concerns. He pointed particularly to the permission to develop advertising and business products, arguing that it remained broad enough to permit activities beyond ordinary advertising functions.
Attorney Matt Topic of Loevy & Loevy raised a similar concern. He said the terms still allow Meta to use collected content to develop business products, a category that could potentially include AI-powered tools.
The concern is not simply whether the policy expressly permits AI model training. It is also whether broad terms such as “develop” could be interpreted to include training or improving AI systems used for commercial purposes.
Meta has not publicly indicated that the revised provision authorizes unrestricted training of its general-purpose AI models on news content. Its statements instead emphasize advertising and business services.
The legal disagreement concerns what the contract could permit, not verified evidence that Meta has begun using the disputed provision to train AI systems on publishers’ websites.
Meta Denies Expanding Its Scraping Practices
Meta vice president of communications Andy Stone disputed suggestions that the revised terms represented a change in the company’s website crawling practices.
The company said it was combining existing agreements, including provisions governing generative AI advertising tools, into a broader commercial policy. Meta described the changes as an effort to simplify its business terms and clarify how its services operate.
A company spokesperson also responded to criticism over website access controls.
“We’ve heard you and want to clarify,” Meta said in a statement reported by TechCrunch.
The company explained that the consolidated crawling provisions concern advertising and business products. It also maintained that website owners can continue communicating their preferences through established technical controls.
Meta said the revisions were not intended to authorize scraping news websites or override restrictions established by content owners.
Its explanation focuses partly on how businesses use information from external websites. For example, advertising services may need accurate product descriptions, availability information and other details to create advertisements or connect users with businesses.
Collecting updated website information can support those functions. The dispute arises because the contractual wording is not restricted to those specific examples.
Why Robots.txt Is Part of the Dispute
One unresolved issue involves robots.txt, a standard file website operators use to communicate which sections of their sites automated crawlers should avoid.
Search engines and other automated services commonly consult these files before collecting website information. Publishers may also use them to discourage crawlers associated with AI development or other activities they do not authorize.
Meta says website owners can continue using robots.txt to communicate their preferences. The company has stated that its updated terms do not create new rights that override those instructions.
However, the revised commercial terms do not explicitly promise to follow robots.txt restrictions.
Benjamin Sobel, a law professor at the University of Wisconsin, explained that robots.txt expresses a website owner’s preferences but does not automatically create an enforceable legal restriction. Whether an automated service must honor those instructions depends on applicable law and other circumstances.
That leaves a distinction between Meta’s public statements about its practices and the permissions described in its commercial agreement.
For publishers, the uncertainty matters because automated collection can involve large amounts of copyrighted material. A website might permit search engine indexing while seeking to restrict other forms of reuse.
The revised terms do not clearly explain how disputes involving those different permissions would be resolved.
News Publishers Raise Copyright Concerns
The controversy has attracted attention from publishing organizations in the United States and Europe.
The European Publishers Council called for European authorities to examine the proposed terms, including their implications for copyright protections, contractual fairness and competition.
The United Kingdom’s News Media Association also urged government intervention, criticizing the possibility that publishers could be required to accept broad website crawling permissions to continue using Meta’s commercial services.
In the United States, the News/Media Alliance raised objections and questioned whether Meta’s approach improperly used its position in digital distribution to secure additional access to publishers’ content.
These organizations argue that news publishers invest substantial resources in reporting, editing and distributing original journalism. They contend that collecting such work for commercial products raises important questions about compensation and control.
The concerns extend beyond large national news organizations. Smaller publishers and independent websites also use Meta platforms to reach readers and promote their work.
Many cannot easily stop using Facebook or Instagram without affecting their audience relationships. That dependence makes changes to business terms especially significant for companies with limited distribution alternatives.
Business Accounts Face an October 30 Deadline
Meta’s updated Commercial Terms apply to a broad range of commercial activities, including advertising, managing business pages and using tools such as Meta Business Suite.
Under the policy, continuing to use covered Meta products for business purposes after the changes take effect constitutes acceptance of the updated agreement.
The terms do not offer businesses a separate option to reject only the website crawling provision while continuing to use the covered commercial services.
Businesses that disagree with the terms may instead stop using Meta products for business purposes. For publishers that depend on those platforms for traffic, advertising or communication with readers, that could carry significant practical consequences.
The policy change also comes amid broader disputes over how technology companies obtain material used in AI development. Publishers and other copyright holders have challenged the use of protected works in AI training, while technology companies have defended various collection practices.
Those wider disputes remain separate from the interpretation of Meta’s revised commercial terms. The present controversy concerns the contractual permissions associated with business accounts and how those permissions may apply to external websites.
Meta’s decision to remove the explicit training language has narrowed the stated purposes of its policy. It has not eliminated disagreements about the scope of automated collection, the meaning of business product development or the enforceability of website access preferences.
With the October 30 effective date approaching, publishers must decide whether to continue using the affected services under the revised agreement. The remaining question is whether Meta’s clarifications will satisfy content owners or prompt further legal and regulatory scrutiny.
IMAGE: Source Own work Author LPS.1 This file is made available under the Creative Commons CC0 1.0 Universal Public Domain Dedication.