How Meta’s AI tools override creative without warning—and what we’re doing about it
Meta has rolled out a growing set of AI tools that can rewrite ad headlines, generate product images, and create videos using your brand assets. These features are promoted as optional, but they are often turned on automatically. In many cases, media buyers do not choose to enable them. Meta applies them silently.
Even if we manually turn them off, they can be reactivated later.
This is causing confusion, especially when a client sees an ad they did not approve. This post explains what is happening, why it keeps happening, and how we’re protecting brand integrity across every account.
What Meta Is Doing
Meta has introduced a set of creative automation tools across its ad platform. These include:
-
– Automatically written headlines and descriptions
-
– AI-generated product photos with new backgrounds or lighting
-
– Auto-created videos from still images and catalog content
-
– Suggested text changes during campaign setup
You will see these most often in Advantage+ Shopping, App, and Catalog campaigns. They may also appear in existing campaigns that were set up before these features were available.
Here’s the key issue: many of these tools are enabled without notification. If we do not check each box during setup, Meta will begin applying changes to your ad creative on its own.
You Didn’t Approve It. Neither Did We.
When you see an ad that doesn’t sound like your brand or looks off, your first question is usually: “Who made this?” In most cases, the answer is Meta.
We have seen live ads with AI-generated headlines that were never submitted. We have found product photos that were altered automatically. These are not manual errors. They are system defaults.
Media buyers are not notified when these features are activated. Clients are not notified when they go live.
Even when we turn them off, Meta may turn them back on later. This happens without a prompt. It may also happen retroactively in older campaigns that were launched before these features existed.
We check for this constantly. But unless every campaign is audited weekly, these updates can slip in.
What This Means for Brand Safety
Most of our clients care deeply about how their brand appears in the world. That includes voice, tone, product presentation, and visual design. AI-generated creative often disregards those standards.
We have seen headlines that contradict messaging strategy. We have seen product images that look mismatched. We have even seen entire ads built from assets we never approved.
This is not just a performance problem. It is a trust problem.
What We’re Doing About It
We now audit every new campaign for AI creative settings. In most cases, we disable:
-
– Auto-generated text and headlines
-
– Product image and video creation
-
– Suggested copy changes
We also retroactively check older campaigns, including those that may have inherited AI tools after the fact.
When necessary, we document every variation running in the account so clients can see what has been published.
We flag anything that does not match your original creative for review and removal.
We are doing this because Meta does not make these changes visible by default.
Why This Blog Post Exists
This post is meant to clarify what happens when an ad appears that you never approved. It is not a mistake by your media buyer. It is a platform-wide shift toward automation that bypasses manual control.
When this happens, the right response is not blame. The right response is awareness and prevention.
We are holding the line. We are keeping your ads clean, consistent, and on-brand. If anything looks off, we will catch it and correct it.
What You Can Do
If you ever see an ad that surprises you, send it our way. We will confirm what happened and fix it.
If you want more visibility into how your creative is being used, we can walk through every active variation with you.
Our priority is keeping your brand safe, not just keeping the ads running.