Meta shifts AI ad risk onto advertisers

Meta is moving AI ad mistakes onto advertisers

Meta is changing the risk equation for paid social. Starting Oct. 30, its updated Commercial Terms say advertisers are responsible for whatever its AI creates or changes, even when Meta switches the feature on. According to ADWEEK, buyers are already seeing the platform rewrite creative without asking first, which turns a product feature into a governance issue. For brands heading into the biggest buying season of the year, that is not a small policy tweak. It is a question of who pays when automation changes the ad, the message or the legal exposure.

The contract now puts the cleanup bill on brands

The new language adds an AI Features section that does not exist in the current version of Meta's terms. Under that section, advertisers who run ads on Meta are stated to be solely responsible for whatever the company's AI features create or change, including features Meta itself switches on. Previously, according to the terms cited by ADWEEK, liability applied when advertisers chose to use Meta's AI tools. Now the language says the AI may make mistakes, may be inaccurate or misleading, and that all risks sit with the advertiser.

A Meta spokesperson did not directly answer whether the new terms make advertisers liable for AI features they never turned on. The spokesperson said advertisers can turn off individual Advantage+ features in Ads Manager at any time. Meta also framed the update as housekeeping, saying it is folding existing generative AI ad terms into its main ad terms. That framing matters less than the practical effect. If a platform can change ad assets by default, and the contract says the advertiser owns the result, then the control point has moved from creative approval to post-change cleanup.

Source: ADWEEK

The problem is already showing up in live campaigns

Bookshop.org said on Threads that an automated account update from Meta opted it into a generative AI feature that added an AI-generated illustration to a human-made ad without the company's knowledge. The bookseller said the ad was enhanced, or ruined, by Meta and has since turned off all of Meta's AI ad features. That example is important because it shows the issue is not theoretical. It is not just about whether a marketer wants AI assistance. It is about whether an account update can change a live campaign before anyone notices.

Other advertisers are describing similar behavior. Exverus told ADWEEK that Meta's AI is adding sitelinks, translating copy, and generating AI images and carousels from a brand's own photos, sometimes on ads where the advertiser did not switch the feature on. Ryan Adams of Exverus said the agency has had to pull live ads to strip out changes that slipped through quality checks. He cited added sitelinks, language translations and AI-generated assets built from the brand's own media. The operational lesson is simple: if the review process assumes the platform will stay hands off, the review process is already broken.

Source: ADWEEK

Peak season makes the risk harder to ignore

The timing matters because Meta is making this change a month before Black Friday and Cyber Monday, when many brands depend on Meta for a large share of holiday demand. David Dweck, president at Go Fish Digital, said Meta knows advertisers are under pressure and may feel forced into the new setup. His comment cuts to the commercial reality: if a platform change lands right before peak season, most teams will not have much room to test alternatives without risking sales volume.

That is why some agencies are not telling clients to pull back in Q4, even while they review the terms with legal teams. At the same time, some buyers are already turning off Meta's default automation across campaigns. Asa Hiken, senior analyst at Emarketer, said automatic generative AI revisions can clash with a brand's strategy and values, and consumers may assume the brand approved the AI output. The result is a classic enterprise risk problem. A platform decision affects revenue, reputation and compliance at the same time, but the brand owns the fallout.

Source: ADWEEK

Meta is also widening the data perimeter

The new terms go beyond ad creative. Meta says businesses authorize it to send web scrapers, bots, robots, spiders, crawlers and user agents to sites associated with their products and services, and to use what it collects to provide, develop, improve, research and maintain advertising and business products. Meta later said the crawling terms were consolidated only for ads and business products, and that they do not change how its web crawlers work. The company also said content owners can use robots.txt to signal their preferences.

For leaders running marketing, e-commerce and legal together, the point is not just whether Meta is scraping more or less. It is that the platform now sits closer to the creative supply chain and the website data layer at the same time. That raises a practical governance question: where does default automation end, where does human approval begin, and who signs off on the risk if the platform changes something the brand did not intend to publish?

Source: ADWEEK

What to do now

This is a good quarter to treat Meta automation as a control issue, not just a media setting. Before the holiday push gets deeper, marketing, legal and operations should agree on what is allowed, what is blocked and who can pull a live ad if an AI change slips through.

• Audit every active Meta account for Advantage+ and other AI features, including settings hidden in collapsed menus.
• Review the updated Commercial Terms with legal and procurement before Oct. 30, and document who owns the risk.
• Add a pre-launch QA step for sitelinks, translations, carousels and any AI-generated image or copy variation.
• Decide which campaigns can tolerate automation and which require manual approval only, especially in Q4.
• Create an escalation rule for pulling or pausing ads fast when Meta changes creative without approval.