Key takeaways
- Meta’s commercial terms change on 20 November 2026 and are accepted by continuing to use its products for business purposes. There is no button and no signature.
- Crawling of your website was already granted in the self-serve ad terms of August 2025. What is new is the reach to any business use and the test credentials.
- Section 5 makes the advertiser “solely responsible” for whatever AI features produce, with no distinction between features the advertiser enabled and features Meta enabled.
- The preview says 20 November in its banner and 30 October in its footer, in English and in Spanish alike. Meta has already changed the text once and may do so again.
What changes in Meta’s commercial terms on 20 November
Meta’s commercial terms change on 20 November 2026 in three places. A new paragraph authorises Meta to crawl the websites associated with your business. A new section, the fifth, covers artificial intelligence features and puts the responsibility on the advertiser. And a list of supplemental terms is folded into the contract, among them the ones governing the pixel and the Conversions API.
There is nothing to sign. The preview Meta has published, read on 27 October 2026, says so in section 7. Anyone who keeps using its products for business or commercial purposes after that date accepts the new terms.
The date has already moved once. Until 23 October it was 30 October. That day a Meta spokesperson told Reuters the company needed more time “to explain to our partners and stakeholders what’s changing and what’s not”.
The preview itself has not fully caught up. The banner says 20 November and the footer still reads “Effective date: 30 October 2026”, in the English version as well as the Spanish one.
The text changed too. Between 17 and 19 October the word “train” disappeared from the preview, as Search Engine Journal and PPC Land reported. Everything below describes the version seen on 27 October 2026, and it is worth rereading on the 20th.
Do I have to accept Meta’s commercial terms, or can I refuse?
You do not have to accept anything expressly. Meta’s commercial terms are accepted through use. Section 7 puts it in one sentence: “By continuing any access or use of any Meta Company Products for business or commercial purposes […] you agree to be bound by them.”
The alternative the text itself offers is to “stop all access or use of our Products for business or commercial purposes”. For a business with a Facebook page, an Instagram profile and live campaigns that is not a real option, and it is better to say so plainly.
Reach matters here. The self-serve ad terms only bound whoever placed an ad order. The commercial terms bind any use for business or commercial purposes, so a Facebook page that never ran an ad is covered too. If you manage social media for your business without spending on advertising, this reaches you all the same.
What you already use gets tied in as well. The new section 8 incorporates by reference the business tools terms, which govern the pixel, the SDK and the Conversions API, in other words all the measurement feeding your campaigns. The data restrictions in section 3 do not change, and data about children under 13 and sensitive categories such as health or finances remain off limits.
Two practical things before the 20th. First, know which legal entity owns each ad account, page, Instagram profile, catalogue and WhatsApp Business number, because that is the entity accepting through use. Second, keep a dated copy of the preview in English and in your own language; the date mismatch in the preview is reason enough.
Section 5 makes you responsible for whatever Meta’s AI changes in your ads
Section 5 is new. The commercial terms in force today, dated 6 March 2024, have no section on artificial intelligence at all. The 20 November preview adds one and allocates responsibility in a specific way. Meta provides the features and the advertiser carries the risk of whatever they produce, including the job of judging whether the result is fit for a commercial purpose.
“Meta may use or make available products and features powered by artificial intelligence, machine learning, generative AI, or other automated products or features that help you understand and use our Meta Company Products and/or optimize, enhance or create Content. You acknowledge and agree that AI features may make mistakes and may not always produce accurate, complete, or suitable results.”
“You are solely responsible for and assume any and all risks associated with such AI features and any resulting Content, including evaluating the suitability of any Content for business or commercial purposes. You must not alter or remove any watermark or content-authenticating metadata generated by AI in connection with Content.”
Meta Commercial Terms, section 5, preview effective 20 November 2026, read on 27 October 2026
The words that matter are “solely responsible” and “any resulting Content”. The text draws no line between a feature you switched on and one Meta switched on by default inside your ad. According to Adweek, which on 22 October published cases of ads altered without permission, the generative AI ad terms now being folded into the commercial terms spoke of using the AI “in your sole discretion”. That phrase is not in the new text.
Meta told Adweek that advertisers can turn off individual Advantage+ features in Ads Manager at any time. It added that the update folds the generative AI terms into the main ones. It did not answer whether an advertiser is responsible for features it never enabled. As of 27 October 2026 that question remains open.
The same section adds two more obligations. Do not alter or remove AI-generated watermarks or content-authenticating metadata, something some creative export workflows do by accident. And answer for “the acts or omissions of any third-party software or service that you use or authorize”, which includes the agency running your account. Check with whoever handles your legal matters how that sits with your contract.
How do I review Advantage+ enhancements before 20 November?
Open every active or scheduled ad, expand the Advantage+ enhancements block and decide one by one which ones stay. Meta’s own Advantage+ creative page acknowledges that some enhancements may be on by default and that you can turn them off whenever you want. Features that use AI show an AI logo under their name, according to that same page.
Previews are not enough. The same page says the versions shown are examples of possible variations and not the full list of what may be served. To see what actually went out, look the ad up in Meta’s Ad Library.
Defaults for new ads are set in the ad account settings. Meta’s Business Help Centre has an article on turning off Advantage+ creative enhancements in Ads Manager. It only opens when you are logged in, so follow its steps from your own account rather than trusting somebody else’s screenshots.
Agencies such as JumpFly and advertisers quoted by Marketing Directo in October 2025 report that enhancements reappear after duplicating an ad or swapping the creative. Meta does not document this. To be safe, repeat the review after duplicating, editing or receiving any account update notice.
If you work through the Marketing API, since version 22.0 you no longer enable or disable the standard enhancements bundle as a whole. It is done feature by feature with enroll_status set to OPT_IN or OPT_OUT, which lets an agency apply one policy across every account at once.
Keep dated screenshots of each ad in the version you approved. If an AI feature changes it, the contract says the responsibility is yours, and the screenshot is the only proof of what you asked for.
According to PPC Land, on 28 October a feature called “product browsing” switches on in ads already running unless the advertiser opts out. There is no confirmation in Meta’s documentation as of 27 October. Meta credits it with 3% more conversions, a vendor figure with no method, like the 18.9% claimed for the GEM model. If you run catalogue ads, check whether the notice has appeared in your account.
What changes in the crawling of your site if you already ran ads?
Little in practice and a fair amount on paper. The self-serve ad terms in force since 22 August 2025 already grant Meta the right to crawl your domains when you place an order. The listed purposes include “research and training”. The 20 November preview moves that permission into the commercial terms, extends it to any business use and adds test credentials.
The new paragraph in section 1 goes beyond crawling. It authorises Meta to “create test credentials to access, index, cache, analyze, collect, retrieve, or crawl the domains, web pages, or URLs […] that are associated with your products or services”. The stated purpose is to “provide, develop, improve, research, and maintain advertising and business products” and to “maintain the integrity and security of Meta Company Products”.
The word “train” is no longer there. As PPC Land reported on 18 October and Search Engine Journal on 19 October, the first preview listed it among the purposes and Meta removed it. Lawyers quoted by SEJ, among them Matt Topic and James Grimmelmann, argue that “develop” is broad enough to cover it; Meta denies this. Those are opinions, and the terms do not settle them.
Test credentials are what worries those lawyers most, because they could open areas behind a login. If your site has a customer area, a paywall or documents reachable by URL, check what can be reached from outside and talk to whoever handles your legal matters. That is a web development and permissions job more than a marketing one.
Meta is not alone in this. Google will switch on image extraction in Performance Max by default on 4 December 2026, including for accounts that had it disabled. Both platforms switch on first and leave the advertiser to opt out afterwards.
Which crawler to block in robots.txt without losing link previews
Meta documents five crawlers with distinct names, and the one serving its advertising products is Meta-ExternalAds. Link previews on Facebook, Instagram and WhatsApp depend on FacebookExternalHit. You can block one without touching the other, because each is declared separately in robots.txt.
| Crawler | What Meta uses it for, per its documentation | What happens if you block it |
|---|---|---|
FacebookExternalHit | Previews of shared links. May bypass robots.txt for security checks | You lose the image and title when your site is shared |
Meta-WebIndexer | Meta AI search index | Your site stops appearing in Meta AI answers |
Meta-ExternalAds | “Improving advertising and other business-related products and services” | The one tied to the purposes in the terms. Previews keep working |
Meta-ExternalAgent | “Training foundation AI models or improving products by indexing content” | No effect on ads or on previews |
Meta-ExternalFetcher | User-initiated requests. May bypass robots.txt | Little effect, because it responds to specific actions by people |
Two warnings. Meta states it caches robots.txt for up to 24 hours, so a change does not take effect immediately. And Meta’s commercial terms do not mention robots.txt anywhere. What exists is a statement from its spokesperson to Reuters, in October 2026, that content owners can still manage access with that file, and a statement is not a clause.
Check your server logs as well. If Meta-ExternalAds already shows up in them, crawling started before 20 November, which fits the permission the self-serve terms already granted. To decide which AI crawlers you let in, use the test we apply in AI SEO: separate the ones that answer searches from the ones that train models.
What changes for a business in the EU or the UK
For a business outside the United States, the contract is with Meta Platforms Ireland Limited and any claim goes to the Irish courts under Irish law. That was already the case in the 2024 terms and does not change. The arbitration clause and its 30-day written opt-out apply only to residents and businesses of the United States. The opt-out countdown PPC Land has calculated is not yours.
Section 6 keeps the reference to the Digital Markets Act. It says nothing in that section limits the rights you have under the DMA and keeps the dispute mechanism for business users under Article 6(12). Meta’s liability cap stays at 100 US dollars or what you paid in the last twelve months, whichever is greater. What grows is the list of risks that land on your side.
European publisher associations EMMA, ENPA, EPC and News Media Europe have complained. On 22 October Spain’s publishers’ body AMI asked the European Commission for urgent interim measures, as reported by EFE. The Commission told Search Engine Journal it will assess the complaint and welcomed the delay; the UK’s CMA will keep scrutinising the text. Neither ordered the delay, and as of 27 October 2026 there is no decision. By way of comparison, Google is testing, according to the trade press, a pilot programme that pays publishers when their content feeds AI answers.
What to check before 20 November
This is an afternoon’s worth of checks, not a project. This is the order we follow when reviewing an account.
- Write down which legal entity owns each ad account, page, Instagram profile, catalogue and WhatsApp Business number.
- Keep a dated copy of the preview terms in English and in your own language, and of the self-serve ad terms in force.
- Open every active or scheduled ad and decide the Advantage+ enhancements one by one. Turn off the ones nobody approved.
- Set the defaults for new ads in the ad account settings, following Meta’s help article.
- Check which Meta crawlers appear in your server logs and decide what to do about Meta-ExternalAds in robots.txt.
- Review which areas of your site behind a login or a paywall are reachable by URL, and raise it with whoever handles your legal matters.
- Put in writing with your agency or freelancer who reviews the AI features and how often.
- Put a note in the calendar to reread the text published on 20 November. Meta has already changed it once.
If you would rather have someone go through this with you ad by ad, we do it as part of campaign management. And on the 20th, reread whatever text Meta leaves published. What this article describes is the version seen on 27 October 2026.
Frequently asked questions
Why did Meta postpone the terms from 30 October to 20 November?
A Meta spokesperson told Reuters on 23 October 2026 the company needed more time “to explain to our partners and stakeholders what’s changing and what’s not”. Neither the European Commission nor the UK’s CMA ordered the delay. Meta presents it as time to explain, although it had already removed the word “train” from the preview in October.
Which law and which courts apply to a business in Europe?
The contract is with Meta Platforms Ireland Limited, and claims go to the courts of Ireland under Irish law, as in the March 2024 terms. Arbitration and its 30-day opt-out only concern residents and businesses of the United States. For specific questions, check with whoever handles your legal matters.
Can I remove the “AI info” label from an ad?
No. Section 5 forbids altering or removing AI-generated watermarks and content-authenticating metadata. According to its help centre, consulted on 27 October 2026, Meta applies the label to images and videos created or significantly edited with its generative AI. Minor edits do not carry it.
Does it affect the pixel and the Conversions API?
Indirectly, yes. The new section 8 incorporates the business tools terms, which govern the pixel, the SDK and the Conversions API. The data restrictions in section 3 do not change, and data about children under 13 and sensitive categories such as health or finances remain off limits.
What about the “product browsing” switch on 28 October?
According to PPC Land, Meta switches that feature on in ads already running on 28 October 2026 unless the advertiser opts out. As of 27 October there is no confirmation in Meta’s documentation and it is not in the terms. The improvement figure attached to it is the vendor’s own, with no method. Check your account for the notice.
