Meta AI Ad Policy 2026: What Advertisers Can Run

Meta allows AI-generated ad creative on Facebook and Instagram in 2026. Images, carousels, slideshows and video made with AI tools are all permitted. The policy is about disclosure and context, not prohibition. The AI Info label is a transparency notice for photorealistic synthetic media, and it does not apply to a product photo you cropped, restyled or arranged with a tool. What does get rejected is a synthetic person presented as a real customer, an altered real person, or a claim the creative cannot support. The rest of this page maps each ad format to what you must disclose.

The confusion is understandable. Between the "Made with AI" label rollout, a mid-2024 rename, and the EU AI Act coming into force, a lot of advertisers assumed Meta was cracking down. It was not. What follows is what the policy actually requires, drawn directly from Meta's official AI labeling announcement and Meta's AI-disclosure requirement in the Transparency Center, so you can run compliant AI-assisted campaigns without the guesswork.

What does Meta's AI content policy actually say?

Meta permits AI-generated advertising. The framework, first announced in April 2024 and matured through 2026, exists to tell viewers when content is AI-generated or has been meaningfully altered by AI, not to block it. There are three mechanisms at work.

Automated detection through industry signals. Meta says it built tools that identify the "AI generated" information carried in the C2PA and IPTC technical standards, so it can label images from Google, OpenAI, Microsoft, Adobe, Midjourney and Shutterstock as those companies add the metadata. When Meta detects these signals in uploaded content, it applies the AI Info label automatically. You take no action.

Self-disclosure. When you upload content created or materially edited with AI tools that do not embed C2PA metadata, Meta expects you to disclose it using the built-in AI disclosure control. Skipping disclosure when your tool does not auto-signal is the most common way advertisers slip out of policy.

Escalation for high-risk content. For content that risks misleading people on matters of public importance, such as deepfakes of real politicians or fabricated news events, Meta may apply a more prominent label or remove the content. Standard product advertising does not fall into this category.

One detail worth remembering: Meta originally called the tag "Made with AI," then renamed it to "AI Info" in July 2024 after photographers complained that lightly edited real photos were being over-labeled. The rename came with a policy softening. Minor AI touch-ups no longer trigger the label the way they briefly did.

When does the "AI Info" label apply to an ad?

The label targets synthetic or heavily manipulated visual content, especially photorealistic depictions of people and events that did not happen. It does not target every use of an AI tool in production.

The label typically applies when:

• Your ad contains a photorealistic AI-generated human, that is, a synthetic person who looks like a real individual.
• Video or audio depicts a real person doing or saying something they did not do.
• An image is a fully AI-generated photorealistic scene presented as if captured by a camera.

The label typically does not apply to:

• AI-assisted copy editing, such as improving a headline or body text.
• AI-optimized targeting, bidding, or budget allocation.
• Minor retouching, color correction, or background cleanup.
• Product photos arranged into a carousel or slideshow with text overlays, because the underlying images are real photographs of a real product.

That last point matters most for how brands actually advertise in 2026, so it deserves its own section.

How does the policy apply to AI-assisted product slideshows and carousel ads?

AI-assisted product slideshows sit in the easiest position under Meta's policy. Here is why. A slideshow or carousel ad built from your own product photography is not a synthetic depiction of a person or an event. The images are real. AI is doing the work around the images, choosing the sequence, writing the hook, sizing the text, and matching a template, not fabricating a photorealistic human or a fake scene.

Meta's photorealistic-manipulation rules are aimed at content that could mislead a viewer about what is real. A carousel that shows your actual product with a caption like "3 reasons this sells out" does not raise that question. Nobody is being deceived about whether the product exists or what it looks like. That is a very different risk profile from an AI avatar claiming to be a real customer.

This is the practical advantage of the slideshow and carousel format for compliance-conscious advertisers. You get scalable AI-generated creative without stepping into the one area, synthetic photorealistic humans, that reliably triggers the prominent label and invites extra scrutiny. If you are new to the format, our guide on how to create TikTok slideshow ads from product images walks through the build, and our roundup of TikTok slideshow ideas for product-based businesses covers angles that convert.

Two caveats keep you clean even here. If you use an AI tool to generate a fully synthetic product scene that never existed as a real photograph, treat that as AI-generated imagery and disclose it. And if your tool embeds C2PA signals on exported images, Meta may surface an AI Info notice automatically, which is fine and requires nothing from you.

Do the same rules apply to AI-generated video ads?

Yes, with one extra trigger. AI video is permitted, and the disclosure logic is the same: real footage assembled with AI is treated lightly, while synthetic content is disclosed. The difference is that AI video ads more often feature a photorealistic AI presenter, and that is the specific case where Meta applies a more visible label rather than tucking the AI Info notice into the post's menu.

A visible label on an AI presenter is not a performance penalty or a policy strike. It is a disclosure, and by 2026 most viewers are used to seeing it. But if avoiding the prominent label is a priority, a product-photo slideshow or carousel reaches the same feeds without a synthetic human on screen. For a broader look at the format economics, see our TikTok UGC ads guide.

What are the account-risk mistakes to avoid?

The label rules are easy. The mistakes that actually get ads rejected or accounts flagged live in advertising standards, which apply to human-made and AI-made creative alike.

Claiming a synthetic person is a real customer. An AI avatar captioned "Sarah, a verified buyer from Denver" crosses into deceptive advertising. Presenting the same avatar as a brand spokesperson delivering a scripted message is fine, because you are not asserting a false identity.

Fabricated testimonials. US Federal Trade Commission rules prohibit fake endorsements, and they apply on Facebook. AI can voice enthusiasm about your product. It cannot pose as a verified purchaser who achieved specific unsubstantiated results.

Impersonating a real individual. Generating an AI likeness of a real person without consent violates Meta's policies and a growing set of state laws. Several US states enacted synthetic-media and likeness rules in 2025. Use purpose-built avatars, not lookalikes of real people.

Skipping disclosure on non-signaling tools. If your generator does not embed C2PA metadata and you produce synthetic imagery, disclose it. This is the single most common avoidable violation.

Unsupported claims. AI labeling and ad claims are separate concerns. Whether a person or a model made the creative, Meta and the FTC still require that performance and health claims be substantiated.

Does your ad need an AI disclosure?

Start at step 1 and stop at the first answer that matches your creative.

Step 1

What is the ad creative actually made of?

Meta Community Standards, Misinformation, read 8 Sept 2026

• Photographs of my real product, sequenced, cropped and captioned

  No actionNo disclosure to make

  Meta labels content when it detects industry standard AI indicators in the file or when the poster discloses AI themselves. A slideshow of your own product photographs carries neither, because nothing in it is a synthetic depiction of a person or an event.

  CineRads builds every slideshow from your own product images and your own copy, which is the branch you just landed on. Get three free from your site URL

  Meta Transparency Center, Labeling AI Content, read 8 Sept 2026

• Real footage or photographs of a real person who agreed to appear, with no AI alteration

  No actionThe AI rules do not engage

  Meta's disclosure requirement is triggered by photorealistic video or realistic-sounding audio that was digitally created or altered. Footage of a real person who agreed to appear, cut and colour-graded but not synthesised, is neither, so a creator ad is an ordinary ad. The moment you use AI to change what that person appears to say or do, start again at the third answer above.

  Meta Community Standards, Misinformation, read 8 Sept 2026

• A photorealistic person, scene, video or voice that AI generated or altered, including a real person whose likeness you are licensed to use

  Continue to step 2

• The image, name or likeness of a real person I have no permission to use

  Do not runThis is an account risk

  Meta restricts or disables accounts that impersonate a person by using their image, name or likeness with the aim to deceive. Use a purpose-built avatar you have the rights to, never a resemblance of a real individual.

  Meta Community Standards, Authentic Identity Representation, read 8 Sept 2026

Step 2

Does the ad ask the viewer to believe that person or scene is real?

Meta Advertising Standards, read 8 Sept 2026

• No. It is plainly brand creative, and no claim depends on it being a real photograph or a real customer

  Continue to step 3

• Yes. It is captioned as a named buyer, a real result, or an event that happened

  Do not runRewrite it before you run it

  Meta's Advertising Standards require ads to comply with its Community Standard on fraud, scams and deceptive practices, and its Misinformation policy says it may label or reject an ad whose photorealistic created or altered content risks materially deceiving people. A synthetic person captioned as a named buyer is that case. A clearly scripted brand message is not, so change the caption, not the creative. Then keep going: the rewritten ad still contains a photorealistic AI person, so it still owes the disclosure in step 3.

  Meta Community Standards, Misinformation, read 8 Sept 2026

Step 3

Is the ad about a social issue, an election or politics?

Meta Newsroom, labeling AI-generated content, read 8 Sept 2026

• No. It sells a product, a service or a subscription

  DiscloseUse Meta's AI disclosure tool

  Meta requires a disclosure through its AI-disclosure tool for photorealistic video or realistic-sounding audio that was digitally created or altered, and says it may apply penalties when that disclosure is skipped. If your generator already writes industry standard indicators into the export, the AI Info label can also appear on its own.

  Meta Community Standards, Misinformation, read 8 Sept 2026

• Yes

  DiscloseDisclose, and get authorised first

  Social issue, electoral and political ads are allowed only after Meta's authorisation process, and Meta stated in April 2024 that since that January advertisers have had to disclose when they digitally create or alter a political or social issue ad in certain cases. This is a stricter route than the standard one above.

  Meta Advertising Standards, read 8 Sept 2026

Does the political ads rule affect product advertising?

No. Meta requires explicit disclosure for AI-generated content in ads about social issues, elections, or politics. That is a separate, stricter policy. If you advertise a physical product, a software tool, or an ecommerce brand, the heightened political requirement does not apply. Your product falls under the standard AI Info disclosure framework, as the Meta Transparency Center's labeling documentation makes clear.

What about the EU AI Act for European campaigns?

The EU AI Act's transparency obligations for AI-generated content became applicable in 2026. Generative AI used for advertising is not classed as high-risk under the Act; it sits in the limited-risk tier, where the core duty is transparency. AI-generated content shown to EU users must be disclosed, and Meta's existing AI Info labeling largely satisfies this for content distributed on its platforms. If you run ads in Germany, France, or other EU markets, enabling Meta's disclosure control on synthetic content and avoiding deceptive claims keeps you compliant under both Meta policy and the Act.

The bottom line

AI-generated ads are fully permitted on Facebook and Instagram in 2026. Product slideshows and carousels built from real photos are the lowest-friction format because they avoid the one area, photorealistic synthetic humans, that triggers prominent labeling. When you generate truly synthetic imagery with a tool that does not embed C2PA signals, enable Meta's disclosure control. Keep your claims substantiated and never present a synthetic person as a real, named customer. Do those things and the policy environment is genuinely favorable to AI-assisted creative.

If you are weighing the cost side of the decision, our breakdown of AI versus human UGC creator costs compares the two, and our library of UGC ad examples that convert shows what strong creative looks like in the feed.

Meta is only one platform: our cross-platform AI disclosure guide compares Meta, TikTok, YouTube, and the EU rules side by side, and the TikTok AI content policy guide covers the TikTok half in the same depth.

Frequently Asked Questions

Are AI-generated ads allowed on Facebook and Instagram in 2026?

Yes. Meta permits AI-generated images, carousels, slideshows, and video in ads. The policy governs disclosure of synthetic or manipulated content, not whether AI creative is allowed.

Do AI-assisted product slideshows get the AI Info label?

Usually not. A slideshow built from real product photos with text overlays is not a synthetic depiction of a person or event, so it generally falls outside the photorealistic-manipulation rules that trigger the label.

When does Meta apply the prominent AI Info label?

Meta applies a more visible label when an ad contains a photorealistic AI-generated human or synthetic video and audio depicting a real person doing something they did not do. Product photography and minor edits do not trigger it.

Do I need to manually disclose AI-generated ads?

Only when your tool does not embed C2PA provenance signals and you are uploading synthetic content. Tools that write the C2PA or IPTC "AI generated" metadata are detected automatically and need no manual action.

Does the political ads disclosure rule apply to my product ads?

No. The stricter written-disclosure requirement applies to ads about social issues, elections, or politics. Ecommerce and product ads use the standard AI Info framework.

Can an AI avatar claim to be a real customer?

No. Presenting a synthetic person as a verified, named customer is deceptive advertising and can violate FTC rules. A clearly scripted brand spokesperson message is acceptable; a fabricated personal testimonial is not.