The news: As Meta pours billions into AI, brands are grappling with faulty AI ad tools that produce confusing, misleading outputs, according to advertisers and agency execs that spoke to Business Insider.

• Ads consultant Jessica Gleim said she routinely sees incorrect outcomes from Meta’s AI advertising creative recommendations. A pajama brand Gleim works with received Meta recommendations for new assets that changed the actual product from a pajama dress to a shirt and pants.
• In another instance, Meta recommended ads centering men for a women-only networking group. Gleim stated that these tools “[are] not usable to help my clients grow their business.”
• Other advertisers noted that, despite some of Meta’s features being off by default, they’ve encountered bugs that turn them on. One executive said that most of their 15 clients using Meta advertising were impacted.
• This comes after outdoor retailer REI ran an Instagram ad that depicted a bike with two sets of handlebars; the brand said Meta auto-enrolled them in AI features that caused this.

Why it matters: Meta’s AI tools promise faster, easier campaign creation—but faulty outputs risk stoking negative consumer sentiment, forcing brands to balance automation’s efficiency with greater creative scrutiny while attempting to avoid negative consumer sentiment.

• Consumers are becoming more hostile to—and less impressed by—AI-made media. Between March 2023 and March 2025, the percentage of US adults describing AI-generated works as “not real art” rose from 22% to 35%; those seeing AI outputs as “fake” grew from 14% to 30%, per Ipsos. For advertisers, that matters because poorly executed AI creative can make brands look careless, inauthentic, or overly reliant on shortcuts.
• 65% of US consumers feel somewhat or very uncomfortable with AI-generated ads, per 2024 EMARKETER/CivicScience research. CivicScience also found that only 12% of US adults surveyed in mid-2025 were more likely to buy products from brands if they knew it used AI in its ads.
• Over 30% of consumers across age groups said that knowing an ad is AI-generated made them less likely to choose a brand, per CivicScience. That was especially true for older demographics, with 42% of consumers over 65 feeling this way.

With attitudes toward AI-generated ads skewing negative, Meta’s mistakes create a bigger problem for brands by making automation feel obvious, sloppy, and unchecked. And even though advertisers aren’t giving up on Meta—Gleim reaffirmed that Meta remains “the best platform” for strong ad results—they are now being forced to apply greater scrutiny to keep AI-enabled campaigns from creating avoidable brand-safety risks.

What marketers should do: As nascent AI tools introduce new brand-safety risks, marketers that pair automation with stronger creative oversight will be best positioned to capture the efficiency gains without damaging consumer trust.

Tighter quality assurance protocols before ad launch will become increasingly important, as well as having clear and robust AI governance built into media buying. Brands should not assume that uploaded assets will always appear as intended—an added review step to assess AI enhancements is now necessary.