Advertisers on Facebook and Instagram will be "solely responsible" for whatever Meta's artificial intelligence features create or change in their ads, including features Meta switches on itself, once the company's revised Commercial Terms take effect on October 30, 2026. The change, reported today by Adweek's Trishla Ostwal, arrives as agencies describe Meta's tools adding sitelinks, translating copy and generating images inside live campaigns without the advertiser's request, and as at least one agency has pulled running ads to strip those changes out.

In Short

Meta has computer tools that can quietly change an ad after a business uploads it, for example by adding a drawing, translating the words or making new pictures from the company's own photos. From October 30, Meta's rulebook for businesses says that if one of those changes is wrong or embarrassing, the business that bought the ad carries the blame, even when the business never asked for the change. Companies that run ads on Facebook and Instagram either have to keep switching these tools off campaign by campaign or accept the risk that comes with them.

What Adweek found in the new terms

Adweek's reporter read the revised Meta Commercial Terms and found a new section titled "AI Features." According to Adweek, the section "does not exist in the current version" of the document. That is consistent with the text currently in force. The UK English edition of the Meta Commercial Terms, effective March 6, 2024, contains no reference to artificial intelligence, generative features or automated changes to ads. Its seven numbered sections cover licences, compliance with law, data restrictions, limits on liability, disputes, updates and conflicts with supplemental terms. The banner at the top of that page states: "We're updating the Meta Commercial Terms. These updates will come into effect on 30 October 2026."

Under the new section, according to Adweek, businesses that run ads on Meta will be "solely responsible" for whatever its AI features create or change. The language reaches features that Meta itself turns on, not only those an advertiser selects. Meta also states in the new terms that its AI features "may make mistakes and may not always produce accurate, complete, or suitable results," and that "any and all risks" sit with the advertiser.

The terms leave one exit. "If you do not agree to the updated terms, you may stop all access or use of our Products for business or commercial purposes," the update reads, according to Adweek. The current text carries a near-identical sentence in Section 6, phrased as "please stop," alongside the rule that "by continuing any access or use of any Meta Products for business or commercial purposes after any notice of an update to these Commercial Terms, you agree to be bound by them." No signature or click is required. A business that keeps running ads on October 31 has accepted the AI Features section.

What changed from the earlier wording

Advertisers were not free of liability before. Adweek quotes earlier wording under which they already carried responsibility for AI output, but only when they chose to use Meta's AI tools:

"You may choose to use the Ad Creative AIs in your sole discretion and that your use of the Ad Creative AIs may result in Output that is digitally created, enhanced or altered and that such Output may be inaccurate, incomplete, misleading, offensive, and/or inappropriate," the terms read, according to Adweek. "You also recognize and agree that you-not Meta-are responsible for your use of any Output, and/or any actions you take in relation to your Ad Content."

Adweek does not name the document that sentence comes from, and it does not appear in the March 2024 Commercial Terms. A Meta spokesperson told Adweek the company is folding its existing generative AI ad terms into its main ad terms, which suggests a separate set of terms covered the "Ad Creative AIs" until now.

The difference sits in two words: "choose" and "discretion." The earlier clause assumed the advertiser opted in. The new section, as Adweek describes it, attaches the same liability whether or not the advertiser did.

The parts of the current contract that already apply

Several provisions carried over from the 2024 version bear on what happens when an AI-altered ad goes wrong. Section 4 caps Meta's aggregate liability arising from business use of its products at "the greater of one hundred dollars (USD 100) or the amount that you have paid us in the past twelve months." Section 5.a requires a business to indemnify Meta against third-party claims "related to your services, actions, content or information on Facebook or other Meta Products," including "reasonable legal fees and costs." And Section 1 extends Meta's licence to content that "you or someone on your behalf (such as your agency that places an ad for you...)" makes available.

None of these sections, in their current form, says whether an image that Meta's model generated from an advertiser's photos counts as the advertiser's content. Adweek's reading is that the AI Features section answers that question in Meta's favour. The full text of that section has not been published in Adweek's report, and PPC Land could not retrieve Meta's preview of the revised terms to check it against the quoted passages.

What advertisers say is happening in live campaigns

The contractual change matters because, according to agencies interviewed by Adweek, the automated edits are already running.

Bookshop.org, the online bookseller, said in a post on Threads that an automated account update from Meta opted it into a generative AI feature, which added an AI-generated illustration to an ad that people had made. "This particular ad was 'enhanced' (or ruined, really) by Meta without our knowledge after an account update opted us in to their new ad features last week," the company wrote. Adweek reports that Bookshop.org has since turned off all of Meta's AI ad features. The Threads post's date is not given in the report.

Exverus, a digital agency, told Adweek that Meta's AI is adding sitelinks, translating copy and producing AI images and carousels from a brand's own photos, sometimes on ads where advertisers did not switch the feature on. Ryan Adams, the agency's paid search and social specialist, described how the settings work in practice.

"Meta auto-applies the enhancements, and many of the options are tucked away in hidden or collapsed menus," Adams said. "If every one of those isn't manually unchecked, the enhancements still apply."

According to Adweek, Exverus has had to pull live ads to remove changes that got through several rounds of quality checks: added sitelinks, language translations, and AI-generated images or carousels built from the brand's media. "Meta puts the responsibility squarely on the advertiser, even when Meta's own tools applied the enhancements," Adams said.

Some buyers now disable the automation by default. "By switching these AI features on by default and shifting 100% of the legal liability onto the advertiser, Meta creates huge operational risks," said Kayla Loecker, director of operations, programmatic at AdCellerant. "That's why we systematically turn off Meta's default AI automation across every campaign-we ensure ads hit the exact target audience and geo requested without unwanted AI modifications, waste, or legal liability."

The "100%" in that statement is Loecker's characterisation, not a figure from the terms.

Who is exposed when the ad is wrong

Asa Hiken, senior analyst at Emarketer, framed the risk around consumers' assumptions. "As we saw with Bookshop, these automatic genAI revisions can be antithetical to brands' strategies and their values, and [can] publish without a brand even knowing it," Hiken told Adweek. "It's the brand that stands to lose here; consumers are unaware of Meta's ad policies and are thus likely to assume that the brand willingly used AI."

That argument has a concrete precedent. In June, REI pulled an Instagram ad for a Van Rysel bicycle after an AI-altered image added a second set of handlebars, illegible frame text and a missing crank arm to a photo from a professional shoot. REI said on June 22, 2026, that Meta had enrolled it in an AI personalization tool without its knowledge. PPC Land reported at the time that Meta had not publicly addressed the enrollment mechanism.

Meta's response

Meta's statements to Adweek did not address the central question. A Meta spokesperson "did not directly address whether the new terms make advertisers liable for AI features they never switched on," according to the report. The spokesperson added that advertisers can turn off individual Advantage+ features in Ads Manager at any time.

Meta described the update as housekeeping, according to Adweek: the spokesperson said the company is consolidating its generative AI ad terms into the main ad terms, but did not comment directly on the AI Features section. Meta did not respond to what Adweek called "multiple specific questions" about advertisers' experiences with its AI features.

The account Meta gives - that each feature can be turned off - and the account Exverus gives - that some options sit in collapsed menus and apply unless every one is unchecked - are not contradictory. Both can be true at once. The practical question for an advertiser is whether a toggle that exists but is easy to miss changes who is liable for the output, and the new terms, on Adweek's reading, say it does not.

The timing problem

The terms take effect four weeks before Black Friday, which falls on November 27 this year. David Dweck, president of marketing agency Go Fish Digital, pointed to that.

"[Meta is] going to opt you into AI features, and they're doing so a month before Black Friday, Cyber Monday-the biggest key season of the year," Dweck told Adweek. "Meta knows that they've got advertisers a bit over a barrel here and are going to be forced into it."

Go Fish Digital's legal team is reviewing the terms and the agency has contacted Meta, according to the report. Yet Dweck said the agency will not tell clients to pull back in the fourth quarter, when brands expect high sales volumes from Meta. "It's kind of the tax of advertising on Meta," he said.

That position reflects Meta's scale. The company reported third-quarter revenue of $64.6 billion, up 26%, on October 21, with advertising contributing $63.06 billion. Adweek puts the Wall Street estimate at about $63.3 billion. The same day, a New Mexico judge ordered Meta to pay $25 billion after a jury found in September that Facebook misled users about how much control they had over their data. Meta is expected to appeal.

October 30 is not the only default date in the window. Meta will turn on product browsing in existing ads on October 28, sending some people who tap an ad to an in-app product page rather than the advertiser's site unless the advertiser opts out. The notice for that change cited an average 3% conversion increase from an experiment across about 117,700 advertisers.

A pattern of defaults

PPC Land has tracked Meta's movement toward default-on creative automation for more than a year. Meta announced on August 25, 2025, that Dynamic Media would be on by default for new Advantage+ catalog ads created through the Marketing API, from September 1, 2025, with full enforcement by October 20, 2025. Developers who wanted the old behaviour had to set the enrollment status to opt out.

By October 2025, Advantage+ creative enhancements activated by default, adjusting backgrounds, brightness, contrast and format, and generating up to 150 variations from an advertiser's assets. Advertisers generally could not preview every combination or see which version a given user received. Meta's claims of a 22% return on ad spend increase for advertisers using creative features drew scepticism from practitioners, including an incrementality test that credited Advantage+ with 17% of the conversions Meta's attribution reported.

The liability shift is the contractual counterpart to that product direction. Each default widens the set of ads that carry machine-made changes. The AI Features section, as Adweek reports it, assigns the outcome of all of them to the buyer.

OpenAI's parallel move

Meta is not alone. On October 20, OpenAI emailed advertisers that it would turn on "creative enhancements" by default for self-serve ChatGPT campaigns, allowing its system to rewrite headlines and descriptions, resize images, change destination URLs and translate ad text. The email gave October 30 as the start date; OpenAI's help page gave October 29. Two of the largest advertising platforms will therefore begin rewriting ads by default within a day of each other, and PPC Land's reporting on the OpenAI change found no explanation of version-level reporting that would show advertisers which variant ran.

The crawling clause, revised

Adweek's report also quotes the crawling provision that drew criticism from publishers this month. Businesses authorise Meta to send "web scrapers, bots, robots, spiders, crawlers, and user agents" to sites associated with their products and services, and to use what Meta collects to "provide, develop, improve, research, and maintain advertising and business products."

That wording differs from the version PPC Land reported on October 16, which let Meta use crawled content to "provide, develop, improve, research, train and maintain the integrity of, the Meta Company Products." The word "train" is absent from Adweek's quotation, and the scope reads "advertising and business products" rather than "Meta Company Products." On October 18, PPC Land found the train sentence still on Meta's preview page a day after The Verge reported its removal. Adweek's quotation indicates the clause has since been rewritten, although the report does not say when.

A Meta spokesperson gave Adweek this statement: "We've heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we've clarified that in the terms themselves. It does not change how Meta uses web crawlers. As always, content owners can use robots.txt to signal their preferences. This is not about scraping news sites, and these updates don't give us new rights that supersede those preferences."

The statement names robots.txt as the control for content owners. It does not say whether the test-credentials language reported on October 16 remains.

Why this matters for advertisers and agencies

The practical burden lands on campaign operations. Under the March 2024 terms, a business already indemnified Meta against claims related to its content and accepted a liability cap of $100 or 12 months of spend. Under the October 30 terms, as reported, the advertiser also carries the risk of content it did not write and may not have seen. An AI-generated translation that mistranslates a price, a sitelink pointing to the wrong page, or a generated image that misrepresents a product is, contractually, the advertiser's error.

Agencies face a second layer. Section 1 of the current terms already treats an agency placing an ad as acting "on your behalf," and the opening paragraph requires anyone using Meta's products for a third party to ensure that party abides by the terms. Exverus and AdCellerant describe manual unchecking across campaigns as the control. Whether that work is billable, and who is accountable when one setting is missed, is now a question for client contracts rather than for Meta.

There is also an asymmetry in disclosure. Meta's terms tell advertisers that AI features "may make mistakes." Consumers who see the ad receive no equivalent notice of who made it, which is Hiken's point: the brand's name is on the ad regardless of which system wrote it.

The open questions are specific. Adweek has not published the full AI Features section. Meta has not said whether liability attaches to features it enabled through an automated account update, as Bookshop.org describes. And it has not said whether version-level reporting will show advertisers which AI-altered variants actually ran, which is the record any advertiser would need to answer a complaint.

Timeline

Summary

Who: Meta, advertisers and agencies running campaigns on Facebook and Instagram, including Bookshop.org, Exverus, Go Fish Digital and AdCellerant, with comment from Emarketer analyst Asa Hiken. The terms were reported by Adweek's Trishla Ostwal.

What: Meta's revised Commercial Terms add an "AI Features" section that, according to Adweek, makes advertisers "solely responsible" for whatever Meta's AI features create or change in their ads, including features Meta switches on. Agencies report Meta's tools already adding sitelinks, translations and AI-generated images to ads without a request, and Exverus has pulled live ads to remove them. Meta says advertisers can turn off individual Advantage+ features and describes the update as consolidating its generative AI ad terms.

When: The terms take effect on October 30, 2026. Adweek published its report today, October 22, 2026, four weeks before Black Friday on November 27.

Where: The Meta Commercial Terms govern business and commercial use of Meta's products worldwide, including advertising on Facebook and Instagram.

Why: The earlier generative AI language put liability on advertisers only when they chose to use Meta's AI tools. The new section removes that condition while Meta continues to switch creative automation on by default, leaving advertisers accountable for content they may not have requested or reviewed.