If Meta's AI keeps changing your ads, how will you know what worked?

What happened

The marketing agency Brainlabs has published a warning about Meta's AI ad tools, according to PPC Land. The post is by Liz DeAngelis, the agency's Managing Director, Programmatic & Media. Her argument is that advertisers who let Meta generate and keep adjusting their ad creative may end up with better campaign numbers and less understanding of why.

Her example is Muse Image, Meta's image model. PPC Land points out that this is a column, not a study. It contains no campaign data, no client results and no response from Meta.

On timing, Brainlabs says Muse Image "is still on its way to Advantage+ creative", which is Meta's set of AI tools that generate and alter ad assets. PPC Land reports that Meta launched Muse Image on 7 July 2026 and said days later that advertisers and agencies would get access within weeks. By PPC Land's count, 125 days have passed since the launch.

DeAngelis does not tell anyone to avoid the tools. She writes that for brands that struggle to produce enough creative, it is worth a test. She would not hand it a whole campaign on day one.

What it means for you

Normally, testing an ad works like this. You change one thing, such as the hook, the product shot, the format, the message or the call to action. You watch what happens, and you carry the lesson into the next piece of work.

DeAngelis says constant change breaks that loop. In her words: "If Meta is generating variations and tweaking creative throughout a campaign, without an approval process or detailed reporting on exactly what changed and why, we might know that performance improved without knowing what actually improved it."

For any advertiser, the practical point is this. If your results get better but you cannot say what improved them, you may not be able to reuse the lesson in your next brief or on another platform such as TikTok or YouTube. PPC Land reports that these lessons may stay locked inside Meta.

Her suggested answer is to test the tools on a small, defined slice of activity, with a clear comparison, so you can tell whether the change really made the difference. It is worth deciding now how you would test it.

Where MarketRead fits

MarketRead runs your ads every day: losers are paused on evidence, winners are given more budget in safe steps, and tired ads are replaced.