Digital Marketing Agency in Pune: What Meta's New AI Ad Disclosure Rules Mean for Your Creative in 2026
If you run Facebook or Instagram ads and any part of the creative — a product photo, a background, a voiceover, an edited video — touched a generative AI tool, Meta may now attach an "AI info" label to that ad automatically, and in India you may also be prompted to self-disclose AI use before the ad goes live. This is not a hypothetical policy footnote. Since June 2026 Meta expanded detection beyond its own AI tools to third-party generative AI tools as well, using provenance signals like C2PA metadata, so ads made with tools such as Midjourney, Adobe Firefly, Canva's AI features or various AI voice generators can now get flagged even when the advertiser never touched Meta's built-in creative AI. As a digital marketing agency in Pune working with retail, D2C, real estate and services clients who have all leaned harder on AI-generated product shots and AI voiceovers over the past year, we have had to build this check into our pre-launch review, and this article walks through what actually triggers the label, what does not, and what an India-based advertiser needs to check before a campaign goes live.
The short version
Meta labels an ad "AI info" in two situations: when it was made or edited with Meta's own generative AI ad tools, or when Meta's detection systems find industry-standard AI provenance metadata (C2PA and similar signals) embedded by a third-party tool. The label shows up on the "About this ad" screen that anyone can open from a live ad, and in some cases directly next to the Sponsored tag. Separately, in regions with specific AI transparency rules — Meta names the EU, California, New York, India and Taiwan — advertisers may be given a self-disclosure option in Ads Manager for AI use that Meta's automated detection might not otherwise catch. None of this is enforced by manual review at scale; it is largely automated pattern matching against metadata and known Meta AI tool outputs, so the rollout is uneven and Meta itself says not every AI-touched ad carries the label yet.
What Actually Triggers the "AI Info" Label
The label is not a general "this ad involved AI in any way" flag. It is specifically about photorealistic or audio-realistic generative content — images, video and audio that could be mistaken for an unaltered photograph, recording or camera capture.
Two independent triggers exist:
1. Meta's own generative AI ad tools. If you used Meta's Advantage+ creative features, background generation, image expansion, or any of Meta's built-in generative tools inside Ads Manager to create or materially edit the creative, the label is applied automatically. You cannot switch this off, because Meta has the generation logs on its own side.
2. Third-party generative AI tools, detected via provenance metadata. Since the June 2026 expansion, when you upload an image or video that was made or edited with an external AI tool — Midjourney, Adobe Firefly, Google's Gemini/Imagen tools, Canva's AI features, ElevenLabs-style AI voice tools, and others that embed Content Credentials — Meta's systems read the C2PA (Coalition for Content Provenance and Authenticity) metadata baked into the file and apply the same label. C2PA is an open, cross-industry standard; Adobe, Google, Microsoft, OpenAI and others already embed it in generated files, and Meta is reading those signals rather than asking the advertiser to declare anything for this specific trigger.
This second point is the one most Indian advertisers are not yet tracking. An agency or in-house team that generates a product background in an AI tool, exports the file, and drops it into Ads Manager without stripping metadata is now exposed to an automatic label — even though nobody at Meta touched the creative and nobody filled in a disclosure form. Whether the label appears depends on whether the exported file still carries that metadata, which itself depends on the export settings of the tool used and whether the file passed through an editor that strips EXIF/metadata (which many compression and resizing tools do).
What Does NOT Trigger the Label
This is the part worth getting right, because over-cautious teams sometimes assume any AI involvement anywhere in the workflow forces disclosure. Meta's own guidance draws a real line between photorealistic media and text.
| Use of AI | Label required? |
|---|---|
| AI-generated or AI-edited photorealistic product image used as the ad's image | Yes, generally |
| AI-generated or AI-edited video (including AI voiceover dubbed over footage) | Yes, generally |
| AI voice clone or synthetic voice narrating the ad | Yes, generally |
| AI-assisted ad copy, headlines or captions (text only) | Generally not required |
| Using ChatGPT/Gemini-style tools to brainstorm angles, draft variations, or write a script that a real actor then reads | Generally not required for the writing itself |
| Meta's Advantage+ automatic placements, budget or bid optimisation (no creative generation) | Not applicable — this is automation, not generative content |
| Stock photography edited only for colour correction, cropping or resizing (no generative fill) | Not applicable |
| Ads about social issues, elections or politics where any image/video/audio was AI-created or edited | Yes, mandatory, regardless of region |
The text exception matters practically: if your only use of AI in a campaign is drafting ad copy variations with a language model, that alone does not put you in the labelling conversation. The trigger is specifically photorealistic imagery, video, or audio that a viewer could otherwise assume is unaltered real footage or a real voice.
Where the Label Actually Shows Up
Two places, and they are not identical in visibility:
- The "About this ad" screen. Every ad on Facebook and Instagram has always had a way to see who paid for it and some metadata about the ad; the AI info label now appears there when applicable. This is a low-visibility, opt-in-to-view location — most users never click into it.
- Next to the Sponsored tag, in some cases. For ads Meta considers higher-sensitivity (this is where the political/social-issue category and some regional rules push the label into more prominent placement), the label can sit directly in the feed next to "Sponsored," which every viewer sees without clicking anything.
For a standard product ad using an AI-generated background, expect the label to live in the "About this ad" screen rather than the main feed placement. For political, social-issue, or election-adjacent advertising, plan for it to be visible up front, because Meta already treats that category as mandatory-disclosure territory independent of the June 2026 expansion.
India-Specific Self-Disclosure: What It Actually Means Day-to-Day
Meta names India specifically as a region where local rules or guidelines carry AI transparency expectations, alongside the EU, California, New York and Taiwan. In practice, for an Indian advertiser this shows up as an additional prompt inside Ads Manager during ad creation: a toggle or checkbox asking you to confirm whether the creative was made or edited with AI, on top of whatever Meta's own automated detection already catches.
The practical difference from the automated label is important. The C2PA-based detection only catches content that carries readable provenance metadata. If your video editor strips metadata during export, or your AI tool of choice does not yet embed C2PA credentials, Meta's automated system may not flag the ad — but the India self-disclosure prompt exists precisely to cover that gap. It shifts the burden from "will Meta's detection catch it" to "did you answer the disclosure question honestly," which is a materially different compliance posture. An agency reviewing client creative needs to treat the self-disclosure prompt as a mandatory step to complete accurately, not a box to skip because the automated label did not fire.
There is not yet a single named Indian statute mandating this the way the EU AI Act creates specific transparency obligations — Meta is responding to a mix of consumer-protection expectations and its own India policy interpretation. Treat the exact legal basis as something to verify with legal counsel if you are running high-stakes categories (finance, health, political); this article is not legal advice, and the regulatory picture in India is still moving.
How This Interacts With ASCI-Style Transparency Expectations
Indian advertisers already navigate the Advertising Standards Council of India's disclosure norms for influencer marketing and, more broadly, its expectation that advertising not mislead consumers about the nature of what they are seeing. The logic behind Meta's AI label is the same logic ASCI applies to influencer disclosures: a viewer should not be deceived about the authenticity of what is put in front of them. If your brand has already built a disclosure habit around influencer content — labelling paid partnerships clearly, as covered in our piece on ASCI influencer disclosure guidelines — extending that same habit to AI-generated ad creative is a small operational step, not a new mindset. The common thread across both is: if a reasonable viewer would assume something is real (a person's genuine endorsement, an unaltered photo) and it is not, say so.
Practically, this means an agency's creative sign-off checklist should now have two disclosure lanes running in parallel: influencer/endorsement authenticity (ASCI) and AI-generated content authenticity (Meta's label plus India self-disclosure). They are different rules triggered by different content, but a client who already accepts one disclosure norm rarely objects to the other once you explain the parallel.
Why This Is a Creative Strategy Question, Not Just a Compliance Checkbox
The label itself does not appear to cause automatic ad rejection or disapproval — Meta's stated approach is disclosure, not prohibition. But three second-order effects are worth planning around:
Trust perception. A viewer who opens "About this ad" and sees "AI info" on a product photo may not care. A viewer who sees it on a testimonial-style video, a "customer review," or an ad implying a real person's experience is more likely to feel misled, independent of whether the label was technically required. The label makes it easier for a skeptical viewer to check, and once checking becomes easy, categories that lean on implied authenticity (testimonials, before/after claims, "real customer" framing) carry more reputational risk when the underlying media is AI-generated.
Category sensitivity. Health, finance, and anything adjacent to political or social-issue content already sit under stricter Meta ad policies. Layering an AI-generated visual or AI voiceover onto an ad in these categories compounds scrutiny — both from Meta's automated systems and from how a viewer reacts to finding out afterward.
Creative workflow discipline. Once a brand adopts AI-generated product shots or AI voiceovers as a standing part of its ad production (which is now common — it is faster and cheaper than photoshoots or studio voice talent for iterative testing), the label is a recurring fact of campaign life, not a one-time issue. That argues for building metadata awareness and disclosure into the production pipeline itself rather than reviewing it ad hoc per campaign.
None of this means avoid AI creative tools. It means treat the disclosure step as part of the production workflow, the same way you already treat brand safety review or ASCI compliance review, rather than something to discover after an ad is live.
A Practical Pre-Launch Checklist for Agencies and In-House Teams
Before a creative asset goes live as a paid Facebook or Instagram ad, run through this:
- Trace the creative's origin. For every image, video and audio element, know whether it came from a camera/real recording, a stock library, or a generative AI tool — and if AI, which tool and whether it embeds C2PA metadata.
- Check for metadata stripping. If the asset passed through a compressor, resizer, or certain video editors after generation, confirm whether C2PA/provenance metadata survived. If it did not, automated detection may miss it — which does not remove your India self-disclosure obligation.
- Answer the Ads Manager AI disclosure prompt accurately when it appears, rather than skipping it or assuming automated detection covers you.
- Flag social-issue, election or political adjacent ads for mandatory disclosure regardless of region — this category already requires it independent of the June 2026 expansion.
- Separate text from media. Confirm the team is not over-disclosing AI-assisted copywriting, which Meta generally does not require a label for — save the disclosure conversation for the categories that actually need it.
- Preview the "About this ad" screen post-launch to confirm what label, if any, actually appears, since automated detection can be inconsistent during a gradual rollout.
- Review testimonial and "real customer" style creative with extra scrutiny — if the imagery, voice or footage is AI-generated, the gap between what a viewer assumes and what is actually true is largest here, and that is where reputational risk concentrates even absent a formal violation.
- Cross-check against ASCI disclosure norms for any ad that also involves an influencer or endorsement element, since the two disclosure regimes can both apply to the same asset.
How This Compares to Google Ads
Google has its own emerging AI-content disclosure framework for advertisers, documented in its ads policy help center, and the general direction across platforms is convergent: photorealistic AI media triggers disclosure expectations, AI-assisted text generally does not, and political/social-issue content gets stricter treatment everywhere. If you run both Google Ads and Meta Ads for the same brand — a common setup for Indian SMBs comparing platform performance, which we covered in more depth in Google Ads vs Meta Ads for Indian SMB lead generation — expect to apply a similar internal creative-review standard across both platforms rather than treating them as separate compliance universes, even though the exact mechanics (labels, self-disclosure prompts, detection methods) differ per platform.
Frequently Asked Questions
Does using AI to write my ad copy require disclosure on Meta?
Generally no. Meta's disclosure framework is built around photorealistic imagery, video and audio — content a viewer could mistake for unaltered real footage or a real voice. AI-assisted text, headlines, or captions fall outside that concern in most cases, though ads about social issues, elections or politics carry stricter rules regardless of format.
Will my ad get rejected if it carries the "AI info" label?
Not inherently. The label is a disclosure mechanism, not a policy violation flag. Ads can run normally with the label present; the risk is reputational (how viewers react to finding out) rather than automatic disapproval, though ads that also violate other Meta ad policies (misleading claims, prohibited categories) can still be rejected on those separate grounds.
How do I know if my AI-generated image will trigger the C2PA detection?
Check whether the AI tool you used embeds C2PA Content Credentials by default (many major tools like Adobe Firefly do), and whether any downstream editing, compression or resizing step strips that metadata before upload. If you are unsure, assume it might be detected and complete the India self-disclosure prompt in Ads Manager regardless, since that closes the gap detection might miss.
Is this the same as Meta's Advantage+ automated ad tools?
No. Advantage+ automation (placement, budget, audience optimisation) is a distinct feature from generative AI creative tools. The AI info label is only about generative content creation or editing — photorealistic images, video, or audio — not about algorithmic delivery optimisation. If you want the fuller picture on how Advantage+ has changed lead-generation campaign defaults, see our piece on Meta's Advantage+ becoming the default for lead generation.
Does this apply retroactively to ads I already ran earlier in 2026?
Meta's own guidance frames the June 2026 third-party detection expansion as applying going forward on ads that are created, edited or newly served, and states the rollout is gradual rather than a hard mandate covering every historical AI-touched ad immediately. If you have active or recently paused campaigns using AI-generated creative, it is worth checking their current "About this ad" screen rather than assuming a past absence of the label will persist.
What to Do Next
If your current or planned Meta Ads creative uses AI-generated product shots, AI voiceovers, or AI-edited video, the practical next step is simple: audit your active campaigns' creative sources this week, confirm whether the India self-disclosure prompt has been answered accurately for each one, and build the origin-tracing step (source 1 in the checklist above) into your standard creative production process rather than treating it as a one-off review. As a digital marketing agency in Pune running Meta Ads for retail, real estate, education and services clients across Maharashtra, we have already had to adjust our own client onboarding and creative review process around this — checking asset provenance before launch, coordinating with clients on when self-disclosure applies, and keeping the distinction between AI-assisted copy (generally fine) and AI-generated photorealistic media (generally needs review) explicit in our briefs. If you want a second set of eyes on an existing campaign's creative or want this built into your ad production workflow going forward, our Meta Ads team can review your current assets, and if you are also managing organic and paid search alongside Meta, our SEO and Google Ads teams coordinate the same creative-authenticity standard across channels. Reach out through our contact page if you would like a walkthrough of what this means for your specific campaigns.
Sources: