Is your AI content being flagged?
AI labels, disclosure, and what marketers need to know
LAST REVIEWED: 4 SEPTEMBER 2026
Yes, Meta and Google can spot some uses of generative AI in ads.
No, that doesn't mean every AI image gets a warning.
And there’s no solid evidence that Instagram automatically buries regular content just because it has an AI label.
The details are more useful than the panic.
What is Meta actually labelling?
Meta has been marking AI-generated and significantly edited content for some time, but how and where those labels appear continues to change.
On 1 June 2026, Meta expanded its AI transparency approach for advertising and began rolling out “About this ad”, a central place for additional information within the three-dot menu on ads.
When Meta's own generative AI advertising tools create or significantly edit an image or video, an “AI info” label may be applied.
But not every use of those tools automatically gets one.
Meta says minor edits that don't significantly alter the image or video, and don't introduce a photorealistic AI-generated person, may not receive an AI label.
Meta is also beginning to detect ads created or edited with third-party AI tools using industry-standard signals.
So the system is more nuanced than:
USED AI = LABEL.
It's also worth separating paid advertising from regular posts on Instagram and Facebook. Meta has transparency systems for both, but the rules and presentation aren't necessarily the same.
How can a platform know AI was used?
It isn't necessarily looking at an image and magically deciding that it looks AI-generated.
One important development is Content Credentials, an emerging standard for recording information about where digital media came from and how it was created or edited.
Think of this as provenance rather than an AI detector.
Information can travel with a file through metadata and other technical signals, allowing platforms and other systems to understand parts of its creation history.
But those signals aren't perfect. Metadata can disappear during normal production, exporting, editing and publishing workflows.
So the presence of provenance information can tell you something useful.
Its absence doesn't prove that AI wasn't involved.
Is Instagram burying AI content?
We can't prove that as a general rule.
This matters because there's a lot of speculation connecting AI labels with reduced reach.
Current evidence supports a much narrower conclusion.
Meta identifies and discloses certain uses of generative AI. It also has separate systems and policies governing what content gets recommended, restricted or removed.
Those are not the same thing.
AI labelling and algorithmic suppression are two different claims.
At the moment, there isn't solid evidence for a blanket rule that a normal brand post loses reach simply because AI was used to make it or because an AI label appears.
If reliable evidence emerges showing a direct distribution penalty, we'll update this page.
Until then, don't confuse disclosure with suppression.
What about Google Ads?
Google made a significant change on 9 July 2026.
It announced expanded AI transparency for advertising, including a global “How this ad was made” section inside My Ad Center across Search, YouTube and Discover.
When advertisers use Google's own generative AI advertising tools, Google can automatically add a disclosure.
When an ad is created elsewhere, Google now provides advertisers with a control to indicate that generative AI was used.
Depending on local requirements, that information may also appear as a label directly on the ad.
The direction is clear: platforms increasingly want information about how an ad was created to travel with the ad.
But a platform disclosure doesn't decide whether an ad is legal, accurate or misleading.
Those are separate questions.
Does Australia require every AI ad to be labelled?
No broad requirement exists in current Australian guidance.
The Australian Government's National AI Centre guidance says:
“You may not need to disclose it every time.”
Instead, it recommends a proportionate approach.
How much did AI contribute?
How much could the content affect the viewer?
The guidance recommends greater transparency when AI-generated content could influence decisions or affect people's rights, safety or trust.
It also says content with low impact and limited AI involvement might need minimal or no disclosure.
That's a more useful test than trying to work out exactly where Photoshop ended and AI began.
Does an AI label mean the content is misleading?
No.
Disclosure and accuracy are separate questions.
An AI-generated background might require disclosure on a particular platform without changing anything material about the product being advertised.
Conversely, an AI alteration that changes an important feature of the product could be misleading regardless of whether it carries an AI label.
The ACCC's rules focus on whether advertising creates a false or misleading impression, including through images and descriptions.
AI disclosure tells people something about how an ad was made. It doesn't decide whether the ad is truthful.
For a broader look at AI, accuracy, synthetic people and advertising production, read:
AI IN ADVERTISING: WHAT MARKETERS NEED TO KNOW →
What do consumers expect?
There is good reason for platforms and advertisers to be thinking about transparency.
A YouGov study published on 20 May 2026 found that 86% of Australians believed brands should clearly disclose when ads were mainly created using AI.
The wording matters.
The research wasn't asking whether every use of generative fill, background extension or AI-assisted retouching requires a label.
It does show meaningful sensitivity when AI plays a substantial role in creating an advertisement.
So when should we disclose it?
There isn't one test that applies to every platform, country and type of advertising.
But there's a useful question to ask:
Would knowing how this was made change what the audience thinks they're seeing?
Removed a lighting stand from the background? Probably low impact.
Generated a fantasy environment? Context dependent.
Created someone who appears to be a genuine customer? Think carefully.
Changed what the actual product looks like or can do? Much bigger problem.
These aren't legal rules. They're a useful way to think about materiality.
The IAB AI Transparency & Disclosure Framework V2, released on 18 August 2026, takes a similar risk and materiality-based approach.
Importantly, the IAB doesn't recommend labelling everything. Its framework specifically considers the problem of unnecessary labelling and “disclosure fatigue”.
That's an important distinction.
Transparency should help someone understand what they're viewing.
It shouldn't become another disclaimer everyone learns to ignore.
Before you publish
Before approving AI-assisted work, check what was generated and what was changed.
Make sure the product still accurately represents what you're selling.
Pay particular attention to synthetic people, testimonials, endorsements and depictions of real events.
Check whether the platform and market where the campaign will run require disclosure.
For significant campaigns, make sure your production partner can explain where key assets came from and what AI was used for.
AI production is moving faster than most policy pages.
That's why this one has a date at the top.
LAST REVIEWED: 4 SEPTEMBER 2026
Want to see what AI could do for your next project?