Meta is facing a fresh question over its upcoming commercial terms: which version of its website-crawling rules will businesses actually accept when the changes take effect on October 30, 2026?
On October 17, Meta said it had clarified language that prompted concerns about publishers’ content being used for AI training. Yet a subsequent review of an official preview found the disputed training provision was still visible. The discrepancy matters for brands, agencies and publishers that operate Facebook Pages, Instagram accounts or Meta advertising tools.
Table of contents
Jump to each section:
- What Meta said it changed
- Why the published preview still raises questions
- Why publishers are concerned about crawling rights
- What marketers should review before October 30
Key Takeaways
- Meta said it removed an AI-training reference from revised terms following criticism, but a later check found the wording still visible in at least one official preview.
- The October 30 terms continue to authorize automated crawling of websites associated with business accounts, even in accounts describing the narrower revision.
- Publishers and marketers should review the specific version of the terms presented to their accounts and document content-access preferences before the effective date.
What Meta said it changed
The initial controversy centered on a clause authorizing Meta and parties acting on its behalf to use scrapers, bots and other automated software on websites associated with a business’s products or services. An additional sentence expressly included training among the permitted uses of the collected content.
After objections from news publishers and legal experts, The Verge reported that Meta had removed the sentence and limited the crawling terms to advertising and business products. Meta spokesperson Daniel Roberts described the intent of the policy and said, “This is not about scraping news sites.” The company also said its actual crawler practices had not changed.
That response addressed the most visible AI-training reference, but left a separate question: whether the text presented to every business user had been updated to match it.
Why the published preview still raises questions
On October 18, PPC Land reported that a UK-English preview of Meta’s October 30 terms still included the original sentence naming training. Its comparison found the wider crawling authorization remained in place too.
A separate check of Meta’s official terms preview for this report also returned language referring to the use of website content to “train” company products. That confirms the disputed wording was still retrievable, though the page response did not establish which locale or account-specific edition would apply to every user.
The evidence supports a narrower conclusion than a universal rollback. Some reporters saw amended wording, while at least one later public version continued to display the original clause. The reason for that mismatch has not been established. Regional differences, a staged rollout or an outdated preview are possibilities, but none is confirmed.
Why publishers are concerned about crawling rights
Even where the training sentence has been removed, the amended language still allows Meta to access, analyze, collect and periodically revisit content from websites linked to business products or services. The scope of those permissions remains contested.
In follow-up reporting by Nieman Lab, Cornell Law professor James Grimmelmann argued that the permission to develop advertising and business products remained broad. Media attorney Matt Topic also questioned how far the term “business product” could extend, including whether it might cover AI-enabled tools.
Meta’s position is different. The company says publishers can use robots.txt to communicate their crawling preferences and that the revised commercial terms do not grant it additional rights overriding those preferences. Legal experts note that robots.txt is a voluntary technical convention rather than, by itself, a guarantee that a crawler will comply.
There is also a commercial tension for publishers. Meta has entered into licensing agreements to bring news into its AI services, while the updated business terms address much broader automated access to websites connected with accounts on its platforms. Those are different mechanisms, and the existence of licensing agreements does not settle the scope of the crawling clause.
What marketers should review before October 30
For marketing teams, the practical issue is the permission a business may be granting through everyday use of Meta’s products. A brand may operate a Facebook Page, advertising account, product catalog and website under different teams, yet the commercial terms apply across business uses.
- Read the exact presented terms. Check the October 30 preview and any account notification rather than relying on a summary or a version cited for another country.
- Document website-access controls. Review robots.txt rules and server logs with the web team, while recognizing that technical preferences and contract language are separate questions.
- Check agency and publisher responsibilities. Identify which Meta assets, domains and external websites are connected to business accounts, including properties managed for clients.
- Escalate unresolved wording. If an account still displays the training clause, record the version and date and seek clarification from Meta or qualified counsel before accepting new commercial terms.
The October 30 effective date now creates a deadline for clarification. Meta has said what the revision is meant to accomplish, but publishers and advertisers need consistency between that explanation and the terms displayed to them.
