The Slop is Coming from Inside the House! The Black-Box Rollout of AI “Enhancements” Inside Digital Ad Platforms
Explore this post further with AI
Imagine our delight then when one of our partners sent us a screenshot of an ad out in the wild on TikTok that had been changed by ad platform AI enhancements. It looked like it was built in the Thomas Kinkade workshop. Unfortunately, as much as people love the Painter of Light(™), his artwork does not adhere to our partner’s brand standards.
If you’ve been working in the digital ad space, especially Meta, Google Ads and TikTok, you’ve seen the evolution of ad platform interfaces over the past few years. New features, new formats, new targeting, AI features. In 2026, we’re reaching critical mass on AI features getting rolled out AND implemented without warning, without consent.
If you’re a solo creator, it’s easy to see how this can be beneficial, how it becomes possible to scale. If you’re an agency working with partners who adhere to strict brand standards, it’s a nightmare.
Welcome to Thunderdome
All bets are off. Anything goes.
Inside Meta, we’ve seen features that we’d toggled off — you’re automagically opted in for most AI features — be magically turned back on during our final QA. We’ve been on video calls, and multiple witnesses have watched this happen. New headlines, remixes of carefully specced content, music added to an ad that already has a soundtrack. Stuff no one asked for.
TikTok is no better, foisting a score of remixed content from CTAs to images to music unless you toggle these features off. At least now you can toggle them off. These features were initially rolled out the same way Robert Irsay rolled the Baltimore Colts out of town: in darkness.
We employ a rigorous QA process for digital campaign launches to double and triple check that settings, budgets, flight dates and all the rest are correct. Lately that includes making sure all the AI features are turned off because we have strict brand standards that our partners require and they pay us to make sure their advertising assets meet those standards.
Probably the most frustrating aspect of intrusive AI features is that they’re always changing. Ad platforms change where you can find the toggle switches or which features are on by default. If you’ve been in Meta ads for the past year or two, you know how this can cause you to question your own competency or even sanity.
As you might expect, this can create difficult conversations internally between the strategic partnership management, creative and ad ops teams. And more importantly, this kind of “AI going rogue” moment can erode trust with our partners.
You’re not alone: the broader pattern
If you’ve experienced these challenges, know that you are not alone. Others in digital advertising have experienced this (and continue to experience this as new features roll out all the time).
If you’re just getting started in Google Ads, new accounts default to auto-applying recommendations unless you can figure out where to turn it off. And Google’s changes keep coming.
The key here is to stay alert and expect more changes in the future.
What do AI enhancements actually do?
AI enhancements on these ad platforms achieve a number of goals. They lower the bar to participation in advertising, which can be a huge boon to small businesses or one-man operations. Another reason is because of competitive pressure; everyone’s doing it. Internally, adoption metrics are important to these ad platforms and their parent companies, so it stands to reason the ad platforms will make these features more prominent and auto-enabled. Finally, every ad place and every campaign spun up is more data that gets fed back into the learning models.
The ad platforms would have you believe these features also result in more efficiency, but in our case the reverse is true as we spend time hunting down and turning off the features or dealing with fallout from auto-enabled hijinks. We’re also led to believe that AI-enhanced ads perform better than anything mere mortals can create. As with most things, this is highly exaggerated. The truth is that AI-enhanced ads sometimes perform better.
How to combat auto-enabled AI features
Here at DR, when we build our campaigns, we have a layer of self-QA, internal ad ops team QA and finally a QA session with someone from the strategic partnership management team. It feels like the most we can do before we hit seriously diminishing returns. For the most part, it works. It’s when enhancements are rolled out after campaigns begin or when older campaigns without enhancements are copied and the enhancements are automatically applied, that the system is susceptible.
Periodic, regular campaign checks, which we already do, are another opportunity to identify campaigns that have had features auto-applied.
The last layer is vigilance on the part of the strategic partnership manager (SPM) and the partner, monitoring their social feeds for anything that looks outside of brand standards.
What we want
These ad platforms’ AI features could be so much better. There could be account-wide settings that allow advertisers who want 100% control over their brand standards to opt out of AI enhancements altogether. If an advertiser could set account-wide settings, they should be persistent, not something that has to be reset for every single campaign, ad group and individual ad. That’s not efficient at all. In addition, AI enhancements should be off by default. It seems like that should go without saying, but here we are.
Ultimately, the advertiser should have control over what tools they use, not the platforms with a conflicting interest. Our responsibility as an agency is to our partners, not to the bottom line of Meta, TikTok or Google.
The bad news is that AI enhancements aren’t going anywhere, and with Meta’s latest overly optimistic AI campaign (one that seems less so when you listen to the soundtrack) and Mark Zuckerberg’s prediction last year regarding fully automated ads by the end of this year, you can be sure we’re going to see more features that are harder to turn off.
If you feel like you’re a tumbleweed blowing through the wild west of AI-enhanced ads, hit us up to talk!

Jonathan Danz
Director of Ad Ops and Analytics