Meta's AI Ads Are Adding Discounts Without Permission. When Does Automation Go Too Far?

AI is supposed to make advertising easier, right? Less manual work, smarter optimisation, better performance and more time for marketers to focus on the creative stuff. That's certainly been the promise behind Meta's growing suite of AI-powered advertising tools.

But what happens when the technology starts making decisions that advertisers haven't actually approved?

Recent reports suggest Meta's automated advertising tools have been pulling promotional offers from retailers' websites and displaying them in Facebook and Instagram ads, even when advertisers had switched promotions off for those campaigns.

For brands managing multiple campaigns, seasonal promotions and different customer offers, that's a pretty significant problem. And it raises a much bigger question about the future of AI-powered advertising: how much control are marketers actually handing over to the platforms?

What's Happening With Meta's AI Ad Tools?

According to reporting from Social Media Today, digital marketing advisor Bram Van der Hallen highlighted an issue where Meta's advertising system was automatically detecting discounts on an advertiser's website and adding them to ads.

The problem? Those discounts weren't necessarily intended for the campaigns they appeared in.

Even when promotional features had been disabled within individual ads, Meta's system could reportedly still surface offers through settings elsewhere in the advertiser's account.

The issue appears to be linked to Commerce Manager, where account-level settings such as Detected offers can allow Meta to identify promotions from a retailer's website and use them across advertising activity.

That means a brand could be running a perfectly normal product campaign, only for Meta to add a promotional message that was intended for a different audience, a separate campaign or a limited-time offer.

And while Meta might see that as an opportunity to make an ad more appealing, the advertiser might see it very differently.

Why an Unapproved Discount Is More Than a Creative Problem

On the surface, an extra discount appearing in an ad might not sound like the end of the world. After all, who doesn't love a bargain?

But for brands, promotional activity is rarely as simple as putting a percentage off next to a product.

Offers are often carefully planned around commercial objectives, margins, stock availability, customer segments and campaign timings. A discount intended to encourage an existing customer to return isn't necessarily something a brand wants to offer to everyone.

Imagine a retailer running a targeted promotion for a particular customer group while simultaneously advertising its full-price collection to new customers.

If an automated system starts pulling that discount into unrelated ads, it could create confusion about which products are eligible, what customers should expect to pay and whether the promotion is actually available to them.

There's also the customer experience to consider. Someone who clicks an ad promising a discount and arrives at checkout to find a different price isn't going to be particularly impressed.

That's not just a potential lost conversion. It's a trust issue.

And for brands investing heavily in paid social, the last thing they need is an automated feature creating inconsistencies between their advertising and the experience customers actually receive.

Meta's AI Automation Is Getting More Powerful, but Is It Getting More Predictable?

Meta has been steadily introducing more automation across its advertising products, from Advantage+ campaigns and AI-generated creative variations to automated placements and performance optimisation.

The idea is that advertisers can provide the assets, objectives and budgets, while Meta's systems identify the combinations most likely to deliver results.

There's a lot of potential in that approach. Automation can help brands test creative at scale, identify opportunities that might otherwise be missed and reduce the manual work involved in managing campaigns.

But the reported discount issue highlights the difference between optimising an advertiser's campaign and changing the offer they're making to customers.

Those aren't quite the same thing.

Changing a background, testing different placements or adjusting how creative is presented is one thing. Introducing a commercial incentive that the advertiser hasn't approved is another.

And as AI-powered advertising tools become more sophisticated, that distinction is going to matter more.

Marketers need to know which decisions are being made automatically, which settings are influencing those decisions and where they can intervene when the system gets something wrong.

Otherwise, the efficiency gained through automation could quickly disappear into the time spent fixing unexpected changes.

The Bigger Issue: Who's Responsible When AI Gets It Wrong?

There's another important consideration here, particularly as Meta continues updating its commercial terms and expanding the role of AI in advertising.

Advertisers remain responsible for ensuring their advertising is accurate, compliant and aligned with the offers they can actually fulfil.

That creates a difficult situation when a platform's automated features introduce changes that weren't explicitly approved by the people managing the campaign.

If an incorrect discount appears, customers are unlikely to care whether it was added by a marketing manager or an algorithm. They'll associate that promise with the brand.

And that's why oversight matters.

AI might be making more of the decisions behind campaign delivery, but the reputational and commercial consequences still sit with the advertiser.

As automation becomes more embedded in paid media, brands and agencies will need to be increasingly clear about who is checking the final output, what changes are permitted and how quickly issues can be identified.

What Should Advertisers Be Checking?

If you're running Meta campaigns, particularly for ecommerce brands with multiple promotions, this is a useful reminder to look beyond the settings visible in individual ads.

Based on the reported issue, there are a few areas worth reviewing:

• Commerce Manager settings: Check whether features such as Detected offers are enabled under offer merchandising settings, and whether Meta can automatically surface promotions found on your website.
• Advantage+ creative enhancements: Review which automated creative features are active and whether they're appropriate for campaigns where pricing, messaging or promotional accuracy is particularly important.
• Active ads and duplicated campaigns: Check the ads customers are actually seeing, not just the original creative and copy submitted during setup.
• Promotional consistency: Make sure the offers appearing in advertising match the products, eligibility criteria and prices customers will find on your website.
• Campaign approval processes: Establish who is responsible for reviewing automated changes, particularly during major promotional periods such as Black Friday and Christmas.

The exact settings available may vary depending on the account and campaign setup, but the principle is the same: don't assume that switching something off in one place means it has been disabled everywhere.

Our Take: AI Should Optimise the Campaign, Not Rewrite the Offer

At Pepper, we're all for using technology to make paid media smarter. AI-powered tools can help brands improve efficiency, test more creative and make better use of their advertising budgets, and there's no question that automation will continue to play a bigger role in how campaigns are managed.

But there's a difference between giving an algorithm room to optimise performance and allowing it to make commercial decisions on behalf of a brand.

A discount isn't just another creative enhancement. It's a promise to the customer.

And if that promise hasn't been approved, the platform shouldn't be making it.

What's particularly frustrating about this situation is that advertisers are increasingly being encouraged to trust automated systems while simultaneously being expected to take responsibility for the results. That only works if brands have clear visibility over what the technology is doing and meaningful control over the decisions that affect their customers.

There's also a wider conversation to be had about the balance between automation and human expertise. The best paid media strategies aren't simply about letting a platform optimise towards a conversion at any cost. They're about understanding the commercial objectives behind the campaign, the audiences being targeted and the wider customer experience.

An algorithm might identify that a discount increases the likelihood of someone clicking an ad, but it doesn't necessarily understand why that discount was created, who it was intended for or what offering it more widely could mean for the business.

That's where experienced marketers still play an essential role.

As AI becomes more deeply integrated into advertising platforms, brands need to be asking more questions about the controls available to them, not fewer. Automation should be helping teams make better decisions, rather than quietly making decisions they never agreed to.

Because ultimately, better advertising isn't just about driving more conversions. It's about driving the right conversions without compromising the brand or the customer experience.

And if Meta wants advertisers to trust its AI tools with more of their campaigns, making sure those tools respect the decisions advertisers have already made feels like a pretty good place to start.