UK advertisers now have until 20 November 2026, not 30 October, before Meta’s new crawling terms apply, after Meta told Reuters on 23 October 2026 that it needed more time to explain them. Meta has cut the word train from the clause, but the licence to crawl sites linked to your ads stands, so the robots.txt work still needs doing.
Key takeaways
- Meta moved the effective date of its updated Commercial Terms and Self-Serve Ad Terms from 30 October to 20 November 2026, a spokesperson told Reuters on 23 October.
- The US English preview of the terms shows 20 November 2026, while the UK English preview still showed 30 October 2026 when we read it on 25 October.
- Both English previews now say Meta may use content from sites associated with a business to provide, develop, improve, research and maintain advertising and business products, with no mention of training.
- The clause still lets Meta use crawlers and test credentials to access, index, cache and collect from those sites, and neither preview mentions robots.txt.
- Reuters reported that publishers and digital advertisers had complained to EU and UK antitrust authorities before the delay, and The Next Web reported that Britain’s competition watchdog has asked Meta to explain the terms.
What moved on 23 October
Eight days ago we set out what Meta’s new crawling terms mean for UK advertisers, with a 30 October deadline. On Friday 23 October a Meta spokesperson told Reuters that Meta was extending the effective date of both sets of terms to 20 November “to allow more time for us to explain to our partners and stakeholders what’s changing and what’s not”. Reuters said publishers and digital advertisers had complained to EU and UK antitrust authorities that the terms would give Meta an “unlimited right to use their content” for its AI services, which Meta rejects.
The Next Web reported that the News/Media Alliance, which represents more than 2,000 US publishers, sent Mark Zuckerberg a letter the day before demanding the terms be withdrawn, and argued that publishers can’t block Meta’s crawlers in practice because their ads depend on them. European publishers asked EU and UK regulators on 21 October to suspend the terms. The Next Web also reported, citing Fast Company, that the disputed wording hasn’t changed since Meta dropped train and that Meta still plans to go ahead.
Which Meta page is right?
The US English preview now carries an effective date of 20 November 2026. The UK English preview still showed 30 October 2026 when we read it on 25 October, two days after the Reuters statement. Meta hasn’t said which page governs a UK account while the two disagree, and a UK legal team shouldn’t guess. Get the date confirmed for your accounts in writing.
What changed in the wording?
The purpose sentence changed. When we read the UK edition on 17 October it said Meta could use content to “provide, develop, improve, research, train and maintain the integrity of” its products. The table compares that with what both English editions say now.
| Part of the clause | UK edition on 17 October | Both English editions on 25 October |
|---|---|---|
| Which sites | Domains, pages or URLs associated with your products or services | The same |
| What Meta may do | Access, index, cache, analyse, collect, retrieve or crawl, plus test credentials | The same |
| What the content is for | Provide, develop, improve, research, train and maintain Meta products | Provide, develop, improve, research and maintain advertising and business products, plus integrity and security |
| Mentions training | Yes, by name | No |
| Mentions robots.txt | No | No |
| Effective date shown | 30 October 2026 | US 20 November 2026, UK still 30 October 2026 |
The UK and US purpose sentences now match, which closes the gap our earlier post flagged. The licence itself didn’t shrink. Meta can still send automated software and test credentials to any site associated with your products, pull updates from them periodically and keep what it collects. “Develop” and “research” remain in the list, and the terms say nothing either way about whether those cover training a model.
Does robots.txt still matter?
It matters as much as it did a week ago. Meta says site owners can still use robots.txt to state their preferences, according to The Next Web, but that promise sits in a press statement rather than in the terms. Meta’s crawler documentation still lists Meta-ExternalAgent as the crawler for uses such as “training foundation AI models”, and Meta-WebIndexer as the one that helps Meta AI cite and link to your content. Meta-ExternalFetcher and FacebookExternalHit may bypass robots.txt in some cases. Meta also notes that its crawlers can cache robots.txt for up to 24 hours, so a rule you add takes a day to bite.
Why the extra weeks help brand teams
Brand and content teams gain the most from the delay. The rights question, whether you can let Meta collect licensed photography, creator content and customer reviews from your product pages, needed legal review that few teams could finish in 13 days. The customer content rights playbook logs what each permission allows, and 26 days is enough to check the licences behind your highest-traffic product pages.
Paid teams get room to stagger their reviews. OpenAI’s switch to ChatGPT Ads creative enhancements by default still lands at the end of October, so ad operations can settle that first and come back to Meta’s terms in November.
Our view at Manual Focus is that the delay changes the deadline and nothing else, so UK brands should finish the plan they started for 30 October. That means blocking Meta-ExternalAgent on domains that carry licensed or customer content, keeping Meta-WebIndexer allowed so Meta AI can cite you, and getting Meta’s answers in writing before 20 November. The trade-off is the one we described in The citation moved off the page, where staying retrievable by AI answers is worth more than any training data a single brand holds back.
Four jobs before 20 November
- Confirm the date for your accounts. Ask your Meta representative which effective date applies to your UK ad accounts and Business Suite, and why the UK preview still shows 30 October. File the reply.
- Ship the robots.txt rules this week. Add the Meta-ExternalAgent Disallow rule on every associated domain where you hold licensed or customer content, confirm Meta-WebIndexer and FacebookExternalHit stay allowed, and check your CDN’s bot settings agree. Doing it now leaves time to watch catalogue and ad delivery before the terms apply.
- Re-brief legal on the new wording. Send them the 25 October text and ask whether “develop” and “research” let your content be used in ways your photography, creator and review licences forbid.
- Write the decision into your brand rules. Record which crawlers you block on which domains and why, so new microsites inherit it. The brand guardrails as code playbook shows how to make that rule checkable.
Open questions for Meta
Meta hasn’t said whether it will change any more of the wording before 20 November or only explain it. It hasn’t said whether a robots.txt block narrows the licence a business grants, or what test credentials can reach behind a login or paywall. We don’t know what Britain’s competition watchdog asked Meta or whether it will act, and the terms still offer no partial opt-out short of stopping business use of Meta’s products. Recheck both previews in the week before 20 November, because any further edit will show there first.
Further reading
- What Meta’s new crawling terms mean for UK advertisers, the original breakdown of the clause and each Meta crawler
- What ChatGPT Ads’ automatic rewrites mean for UK advertisers, on the other platform default changing this month
- Customer content rights pipeline, for checking which content you may let others collect
- What the CMA’s AI choice screen plan means for UK marketers, on the UK regulator’s wider work on AI and search
- The citation moved off the page, on why being cited by AI answers matters more than any one crawler
If you want help setting a crawler policy across your brand’s sites, our services page explains how we work.
Frequently asked questions
When do Meta's new Commercial Terms take effect?
Meta told Reuters on 23 October 2026 that its updated Commercial Terms and Self-Serve Ad Terms will take effect on 20 November 2026 instead of 30 October. The US English preview now shows 20 November. The UK English preview still showed 30 October on 25 October, so UK businesses should ask Meta in writing which date applies to their accounts.
Did Meta remove AI training from its Commercial Terms?
Meta removed the word train from the crawling clause after publishers objected, according to The Next Web. Both English previews now say Meta may use content from sites associated with a business to provide, develop, improve, research and maintain advertising and business products. The terms neither name nor rule out training, and the right to crawl and use test credentials is unchanged.
Do I still need to change robots.txt for Meta's crawlers?
Yes. Meta's revised Commercial Terms still let it crawl, index and collect from domains associated with a business, and neither English preview mentions robots.txt. Meta says site owners can still state preferences there, and its crawler documentation separates Meta-ExternalAgent, which gathers content for uses such as training foundation models, from Meta-WebIndexer, which helps Meta AI cite and link to sites.
Why did Meta delay its new terms to 20 November?
A Meta spokesperson told Reuters the delay allows more time to explain to partners and stakeholders what is changing and what is not. Reuters reported that publishers and digital advertisers had complained to EU and UK antitrust authorities that the terms gave Meta an unlimited right to use their content for its AI services, a claim Meta rejects.
Sources
- Meta Commercial Terms, US English edition effective 20 November 2026, Meta
- Meta Commercial Terms, UK English edition preview, Meta
- Meta Web Crawlers, Meta for Developers
- Meta to delay new commercial terms criticised by publishers, advertisers, Reuters
- Meta delays its website-crawling terms after a publisher backlash, The Next Web
From the podcast
James and Hannah update last week’s episode on Meta’s website-crawling terms. Meta told Reuters on 23 October 2026 that the new Commercial Terms and Self-Serve Ad Terms will take effect on 20 November instead of 30 October, though the UK English preview still showed 30 October on 25 October. Meta has removed the word train, so the UK and US editions now match, but the licence to crawl and use test credentials on sites linked to your ads is unchanged and the terms still don’t mention robots.txt. They cover why the robots.txt plan stands, who gains most from the extra weeks, and three jobs to hand to AI.
Get AI on this
Paste these into whichever AI assistant you use. Swap the bracketed parts for your own details. The first and third need an assistant that can open web pages, and the third also needs one that can run on a schedule.
1. Check robots.txt for Meta’s crawlers
I run marketing for [brand] at [main website], selling mainly in [market].
Meta's new Commercial Terms let it crawl any site associated with our products
or services. We want to block Meta-ExternalAgent, which Meta says crawls for
uses such as training foundation AI models, on sites that carry licensed or
customer content, while keeping Meta-WebIndexer and FacebookExternalHit
allowed.
Here are the domains linked from our Meta ads, catalogues and Business Suite:
[paste list]. For each, tell me which hold licensed photography, creator
content or customer reviews: [mark them, or say "unknown"].
Open each domain's /robots.txt and give me a table with the columns: Domain |
Licensed or customer content (yes, no, unknown) | Meta-ExternalAgent
(blocked, allowed, not mentioned) | Meta-WebIndexer | FacebookExternalHit |
Lines to add or change.
Then write the exact robots.txt lines for each domain that needs a change. If
you couldn't open a robots.txt file, say so in that row rather than guessing.2. Re-brief legal on the new wording
I run marketing for [brand]. Our legal team reviewed Meta's new Commercial
Terms against our content licences before the original 30 October date, and
Meta has since changed the wording and moved the date to 20 November 2026.
Here is the UK clause as it read on 17 October: [paste old wording]. Here is
the clause as it reads now: [paste current wording from Meta's UK preview].
Here are the types of content on our product pages and the licences behind
them: [list, e.g. studio photography, creator content, customer reviews].
Write a one-page note for legal with three sections: what changed in the
wording, what stayed the same, and the questions to put to Meta about whether
"develop" and "research" allow uses our licences forbid. Quote the clause
exactly where you rely on it. Don't give a legal conclusion, and say where the
wording is unclear rather than interpreting it.3. Watch Meta’s terms until 20 November
Set this up as a routine that runs every three days until 20 November 2026.
Open Meta's Commercial Terms previews in UK English
(en-gb.facebook.com/legal/commercial_terms_preview) and US English
(www.facebook.com/legal/commercial_terms_preview). On 25 October the US page
showed an effective date of 20 November 2026 and the UK page showed
30 October 2026, and both said Meta may use content from associated sites to
"provide, develop, improve, research and maintain advertising and business
products".
Each run, record for each page: the effective date shown, the purpose sentence
of the crawling clause word for word, and whether the page mentions
robots.txt. Compare with the last run and tell me in three lines or fewer what
changed, or "no change". If a page won't open, say so rather than reporting no
change.Read the episode transcript
James: A week ago we told you Meta's new crawling terms would switch on for every UK advertiser on the thirtieth of October. They won't, at least not then. Meta's pushed the date back to the twentieth of November and changed one important word, so this is the update. I'm James Vickers, this is Manual Focus, and Hannah's back to go through what moved.
Hannah: Hi James. So on Friday the twenty-third of October, a Meta spokesperson told Reuters it was extending the effective date of its updated Commercial Terms and Self-Serve Ad Terms to the twentieth of November. The reason they gave was more time to explain to partners and stakeholders what's changing and what's not.
James: What's changing and what's not. That's doing a lot of work.
Hannah: It is. Reuters also reported that publishers and digital advertisers had complained to EU and UK antitrust authorities, saying the terms gave Meta an unlimited right to use their content for its AI services. Meta rejects that.
James: And it wasn't only the regulators.
Hannah: No. The Next Web reported that the News/Media Alliance, which represents more than two thousand US publishers, wrote to Mark Zuckerberg the day before demanding the terms be withdrawn. And it says Britain's competition watchdog has asked Meta to explain them.
James: Here's the bit that confuses things. Isn't the UK page still showing the thirtieth?
Hannah: It does. We read both English previews on the twenty-fifth. The US edition shows the twentieth of November. The UK edition still showed the thirtieth of October, two days after the Reuters statement. Meta hasn't said which page governs a UK account while they disagree.
James: So the first job for a UK legal team is boring but real. Get the date confirmed for your own accounts, in writing.
Hannah: Yes. Don't guess.
James: Now the word. When we covered this on the seventeenth, the headline for UK listeners was that the UK edition said Meta could use your content to train its products, and the US edition didn't.
Hannah: That's gone. Meta removed the word train after publishers objected. Both English editions now say Meta may use the content to provide, develop, improve, research and maintain advertising and business products, plus integrity and security. So the UK and US purpose sentences match now, which closes the gap we flagged a week ago.
James: So that's good news.
Hannah: Partly. The licence itself didn't shrink. Meta can still use crawlers, scrapers, bots, and test credentials, to access, index, cache and collect from any site associated with your products. It can still pull updates periodically. And develop and research are still in the list. The terms don't say either way whether those cover training a model.
James: So they took out the word that made headlines and left the permission behind it.
Hannah: That's a fair reading of the text. And The Next Web, citing Fast Company, says the disputed wording hasn't changed since train came out, and Meta still plans to go ahead.
James: Then robots.txt. Does that advice change?
Hannah: Not at all. Robots.txt is the file on your site that tells crawlers what they can visit. Meta says, according to The Next Web, that site owners can still use it to state their preferences, but that promise is in a press statement, not in the terms. Neither English edition mentions robots.txt.
James: Remind people which crawler is which.
Hannah: Meta's own documentation lists Meta-ExternalAgent as the crawler for things like training foundation AI models. Meta-WebIndexer is the one that helps Meta AI cite and link to your content. The other two, the fetcher for user requests and the Facebook link previewer, may bypass robots.txt in some cases. And Meta says its crawlers can cache your robots file for up to twenty-four hours, so any rule takes a day to bite.
James: Our view at Manual Focus is that the delay changes the deadline and nothing else. Finish the plan you started for the thirtieth. Block ExternalAgent on domains with licensed or customer content, keep WebIndexer allowed so Meta AI can still cite you, and get Meta's answers in writing before the twentieth of November.
Hannah: And the extra weeks help brand teams most. Whether you can let Meta collect licensed photography, creator content and reviews off your product pages needs legal review, and thirteen days wasn't much. Twenty-six is enough to check the licences behind your busiest pages.
James: And paid teams can deal with the OpenAI ads default we covered on the twenty-first first, then come back to Meta in November.
Hannah: One deadline at a time.
James: Right. Get AI on this. First job, the robots.txt check across every domain linked to your Meta ads.
Hannah: Give it the list of domains from your Meta ads, catalogues and Business Suite. It reads each site's robots.txt and tells you, site by site, whether Meta-ExternalAgent is blocked and whether WebIndexer and the link previewer are still allowed, then drafts the lines you're missing. Someone on the web team decides and deploys them.
James: Second, get AI on the legal re-brief.
Hannah: Give it the old UK wording from our first piece, the new wording and your licence types, photography, creator and reviews. It hands back a one-page note for legal showing what changed, what didn't, and the questions to ask about develop and research. Legal makes the call.
James: And third, set AI on a watch. Have it check both of Meta's English previews every few days until the twentieth, and tell you if the date or the wording moves.
Hannah: That needs an assistant that can open web pages on a schedule. It compares the date and the purpose sentence with the last check and flags any change.
James: The prompts are in the show notes.
Hannah: What we don't know yet. Whether Meta will change any more wording before the twentieth or only explain it. Whether a robots.txt block narrows the licence at all. What test credentials can reach behind a login or a paywall. What Britain's competition watchdog asked, and whether it'll act. And there's still no partial opt-out short of stopping business use of Meta.
James: So, same plan, new date. Confirm which date applies to you and ship the robots rules this week. The full article, with the old and new wording side by side, is on manual-focus dot co dot uk. Follow the show wherever you listen and the next update arrives on its own, and if you want a hand with a crawler policy, book a free working session at manual-focus dot co dot uk. Thanks, Hannah.
Hannah: Happy to, James. See you on the twentieth, probably.