Meta’s new Commercial Terms now take effect on November 20, 2026, not October 30. A Meta spokesperson told Reuters on October 23 that the company is “extending the effective date of our updated Commercial Terms and Self-Serve Ad Terms to 20 November to allow more time for us to explain to our partners and stakeholders what’s changing and what’s not” (PPC Land). If you read my October 22 roundup, that post used the original October 30 date; this is the update.
The delay changes the date, not the deal. The Meta Commercial Terms still let Meta crawl the websites “associated with your products or services,” still let it create test credentials, and still count continued use as acceptance. What you get is three more weeks to decide what Meta’s crawlers can reach on your site. Below is what moved, what didn’t, and the robots.txt setup that makes sense for a brand that advertises on Meta.
Key takeaways
- New date: November 20, 2026 for both the Commercial Terms and the Self-Serve Ad Terms, per Meta’s statement to Reuters. Meta announced no change to the wording.
- Meta’s own preview page lagged. On October 23 and 24, the UK English preview still showed October 30 (Relevant Audience). Check the banner yourself before relying on either date.
- The crawling clause survives, minus the word “train”. Its stated purpose is now “advertising and business products”. Robots.txt is promised in Meta’s statements, not written into the contract.
- Acceptance is automatic. Using any Meta business product after the effective date binds you. There’s no partial consent.
- Your control is per crawler. Block Meta-ExternalAgent if you don’t want training use; keep facebookexternalhit open or your link previews break.
What changed: Meta Commercial Terms timeline
| Date (2026) | What happened |
|---|---|
| Early October | Meta notifies businesses of updated Commercial Terms, effective October 30 |
| October 16 | Coverage of a new Section 1 letting Meta crawl business websites and use content to train AI; Meta spokesperson Andy Stone calls the reports “false” |
| October 17 | The Verge reports Meta removed the “train” sentence (via PPC Land) |
| October 22 | WAN-IFRA sends Meta a letter asking it to suspend the crawling provision for news publishers (The Keyword) |
| October 23 | Meta tells Reuters the effective date moves to November 20; the European Commission welcomes the delay |
| October 28 | Product browsing switches on in existing ads; not reported as delayed |
| November 20 | New effective date; continued business use means acceptance |
Sources for the October 16 to 23 entries: PPC Land’s original report and its follow-up on the delay.
Why Meta delayed the terms
Meta’s stated reason is explanation, not revision. The pressure came from publishers. According to PPC Land, publishers and digital advertisers complained to EU and UK antitrust authorities that the contract would give Meta an unlimited right to use their content for AI. The Keyword reports that the UK’s News Media Association asked the Competition and Markets Authority to secure a pause, and that European press associations EMMA, ENPA, EPC and NME asked the European Commission for interim measures.
Not everyone is satisfied. The European Publishers Council wants the terms suspended while regulators investigate. Its executive director, Angela Mills Wade, told Reuters: “Maintaining a business page or advertising a subscription should not confer general rights over a publisher’s website and archive.” The NMA’s chief executive, Theo Bamber, called the delay “a pause not a solution” on LinkedIn.
There are more dates to watch before November 20. PPC Land reports that the US-based News/Media Alliance gave Meta until October 26 to respond publicly to its open letter. The European Commission, in the same response that welcomed the delay, said it had sent requests for information to AI-model providers that haven’t yet published summaries of their training data, an obligation under the EU AI Act. Reuters didn’t say whether Meta was one of them. In the UK, Bamber asked the CMA to be ready to begin a Strategic Market Status designation if Meta refuses to drop the contested elements. That is an advocacy request from a party to the dispute, not a CMA proceeding.
Common Thread Collective’s read for ecommerce brands is the practical one: “The deadline moved, but the terms didn’t” (Common Thread Collective). Tech Insider makes the same point: Meta “did not withdraw the terms. It bought time to explain them” (Tech Insider).
What the crawling clause says now
The current preview, as quoted by PPC Land on October 23, authorises Meta “(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, analyse, collect, retrieve or crawl the domains, web pages or URLs (and any content available through them) that are associated with your products or services,” and to “periodically pull updates” from them. The Keyword quotes the stated purpose: to “provide, develop, improve, research, and maintain advertising and business products.”
What was removed
The original version included a sentence letting Meta use content from those domains to “provide, develop, improve, research, train” its products. That free-standing “train” sentence is gone from the UK English preview as of October 23. Critics aren’t reassured: according to Single Grain, a media and IP attorney told Nieman Lab he thinks “develop” is broad enough to include “train” (Single Grain).
What stayed
- Scope: any URL “associated with your products or services.” For an ecommerce brand, that’s your product pages, blog and landing pages.
- Test credentials: the right to create accounts to get past registration walls survives intact, per PPC Land.
- No robots.txt promise in the text. Meta’s spokesperson said content owners can still use robots.txt to signal preferences, but “the preview text published on facebook.com does not mention robots.txt at all” (PPC Land).
- Other clauses: Section 5 still makes the business “solely responsible” for what Meta’s AI features create, Section 6a still requires you to indemnify Meta, and Section 9g still lets Meta change or end features without notice.
How acceptance works
Section 7 reads: “By continuing any access or use of any Meta Company Products for business or commercial purposes after any notice of an update to these Commercial Terms, you agree to be bound by them.” There is no form to sign. PPC Land’s reading is that there’s no partial consent either: you can’t accept for your Facebook Page and refuse for your website. Common Thread Collective confirms there’s no formal opt-out for advertisers who want to stay on the platform.
For US businesses, Section 6c lets you opt out of arbitration and the class action waiver by writing to Meta within 30 days of first acceptance. If acceptance happens on November 20, PPC Land calculates that window closes around December 20, 2026.
Meta’s crawlers: which one does what
Meta’s developer page for web crawlers returned an error when it was checked for this post, so this table relies on sources that quote it: PPC Land’s crawler explainers for Meta-ExternalAgent and Meta-WebIndexer, Single Grain and Playwire.
| Crawler (robots.txt token) | What it does | Can it bypass robots.txt? | If you block it |
|---|---|---|---|
| meta-externalagent | Trains foundation models (Meta AI, Llama) and indexes content directly | No documented exemption | Stops training crawls going forward; no effect on link previews |
| meta-webindexer | Improves Meta AI search results | Not listed as exempt | Meta says allowing it helps Meta AI cite and link to your content |
| facebookexternalhit | Builds link previews for shares on Facebook, Instagram, Messenger | May bypass during security or integrity checks | Breaks your link previews |
| Meta-ExternalFetcher | Fetches single links when a user asks | May bypass, because a user requested it | Not documented in these sources |
| Meta-ExternalAds | Crawls for advertising and business products | Not stated in these sources | Meta doesn’t document the effect on ad delivery |
One gap makes this harder. PPC Land notes that Meta-ExternalAgent covers training and indexing under one name, so you can’t say “index me, don’t train on me” with that token. Google separates Googlebot from Google-Extended; Meta doesn’t. And Single Grain points out that Meta hasn’t said which user agent will do the crawling authorised by the new terms.
The robots.txt setup most brands need
For most brands that advertise on Meta, Single Grain’s default is the sensible one: block Meta-ExternalAgent if training is your concern, and keep facebookexternalhit and Meta-WebIndexer open. Playwire gives the same core rule:
User-agent: meta-externalagent
Disallow: /Leave facebookexternalhit alone. 51Degrees classifies it as a “Preview” crawler, separate from Meta’s AI agents (51Degrees). Playwire’s table is blunt about what blocking it does: it “breaks link previews on all Meta platforms” (Playwire). For an advertiser whose organic posts and customer shares drive traffic, that’s a real cost.
Don’t touch Meta-ExternalAds on landing pages without testing. Single Grain notes Meta doesn’t document what blocking it does to ad delivery, and suggests testing on a single folder first.
Things robots.txt can’t do
- It’s a request, not a lock. Google’s own documentation says robots.txt “cannot enforce crawler behavior” and recommends other methods, like password protection, to keep content secure (Google Search Central). Playwire suggests firewall rules or .htaccess when compliance is in doubt.
- It isn’t instant. The robots.txt standard says crawlers “SHOULD NOT use the cached version for more than 24 hours” (RFC 9309), and PPC Land reports Meta says changes can take up to 24 hours to register.
- It doesn’t undo past collection. PPC Land notes a 2026 rule has no effect on text collected in 2024.
- Tokens are case-insensitive. RFC 9309 says crawlers “MUST use case-insensitive matching,” so
meta-externalagentandMeta-ExternalAgentboth work.
How to check it worked
Meta doesn’t publish a JSON list of crawler IP ranges. PPC Land says Meta directs operators to confirm via whois that a request comes from autonomous system 32934, Meta’s network. In practice: search your server or CDN logs for the user agent string meta-externalagent, wait at least 24 hours after the change, and see whether requests to disallowed paths stop. Third-party directories such as AI Crawler Check list the tokens Meta uses if you need the full set.
Should you block Meta’s AI crawler at all?
For most ecommerce brands, the answer is: block training, allow everything else. Product pages and buying guides are what you want found, and Meta-WebIndexer is the path to citations in Meta AI. Training is a different exchange, and the new terms give you no contractual assurance about it.
There is a traffic trade-off to know about. PPC Land reports that publishers who blocked AI crawlers lost about 7% of weekly site traffic within six weeks, according to an updated Rutgers and Wharton study. That figure is about blocking AI crawlers broadly, which is why blocking only the training token makes more sense than a blanket Meta block.
Publishers face a harder call, because the clause reaches whole sites and archives. WAN-IFRA asked Meta to write robots.txt into the terms and to let publishers decline the clause without losing access to Meta’s business tools (The Keyword). Until Meta answers, robots.txt plus server-level rules is the only control you have.
If you want your site found in AI answers as well as Google, my free book Get Found, Get Clicked covers search visibility, and my Meta BFCM case study covers a peak-season Meta campaign.
Checklist before November 20
| Task | Why |
|---|---|
| Open the preview at facebook.com/legal/commercial_terms_preview and note the date in the banner and footer | Meta’s statement and its page disagreed on October 23 and 24 |
| Add a meta-externalagent group to robots.txt if you don’t want training crawls | The only signal Meta has said it respects |
| Confirm facebookexternalhit is not blocked | Keeps link previews working |
| Check logs 24+ hours later for meta-externalagent hits on disallowed paths | Robots.txt is a request, not enforcement |
| List Advantage+ creative and other AI options switched on in each ad account | Section 5 makes you solely responsible for their output |
| US businesses: decide on the arbitration opt-out and diarise ~December 20 | 30-day window after first acceptance |
| Record when you reviewed the terms and what you changed | Common Thread Collective recommends documenting the review |
Want this applied to your accounts? Get in touch with Zain. I reply on WhatsApp, email and LinkedIn.
FAQ
When do Meta’s new Commercial Terms take effect?
November 20, 2026, according to a Meta spokesperson’s statement to Reuters on October 23. The date was originally October 30. Meta’s preview page still showed October 30 on October 24, so check the page directly.
Can I opt out of Meta’s new terms?
Not while still using Meta for business. Continuing to use any Meta business product after the effective date counts as acceptance. US businesses can separately opt out of arbitration in writing within 30 days of first acceptance.
How do I block Meta’s AI crawler?
Add a robots.txt group with User-agent: meta-externalagent and Disallow: /. Changes can take up to 24 hours to apply. Don’t block facebookexternalhit unless you want to lose link previews.
Does blocking meta-externalagent affect my Facebook link previews?
No. Link previews come from facebookexternalhit, a separate crawler. Blocking meta-externalagent stops training crawls without affecting previews, according to Playwire and Single Grain.
Did Meta remove AI training from the terms?
It removed the sentence that used the word “train.” The crawling right itself remains, with the purpose now described as advertising and business products. Critics argue “develop” can still cover training.
Want this applied to your accounts? Get in touch with Zain. I reply on WhatsApp, email and LinkedIn.
Sources
- PPC Land: Meta delays website-crawling terms to Nov 20 after publisher complaints, Oct 23, 2026
- PPC Land: Meta’s new terms let it crawl business websites to train AI from Oct 30, Oct 16, 2026
- PPC Land: Explaining Meta-ExternalAgent, Aug 29, 2026
- PPC Land: Explaining Meta-WebIndexer, Aug 29, 2026
- Relevant Audience: Meta delays its new Commercial Terms to 20 November, Oct 24, 2026
- Common Thread Collective: Meta Delays Its New Ad Terms Deadline to November 20, Oct 23, 2026
- The Keyword: Meta Delays the Terms That Let It Crawl Business Websites, Oct 23, 2026
- Tech Insider: Meta Commercial Terms Delayed to Nov 20, 2026, Oct 23, 2026
- Single Grain: Meta-ExternalAgent Explained: Should You Block Meta’s AI Crawler?, Oct 23, 2026
- Playwire: How to Block Meta AI From Accessing Your Website Content
- Google Search Central: Robots.txt Introduction and Guide
- IETF RFC 9309: Robots Exclusion Protocol (2022)
- AI Crawler Check: Meta Crawlers: User Agents & robots.txt Rules
- 51Degrees: Meta’s robots and crawlers
