AI Content Disclosure Compliance Checklist: How to Label AI-Generated Marketing in 2026

AI content disclosure is not a single rule — it is a stack of overlapping obligations from platforms, regulators, and trade groups, and the stack is updating multiple times per year. This guide breaks it into a single linear pre-publish checklist so a marketing or creative-ops team can apply it once per asset and have confidence the asset is shippable across every active channel. Treat the checklist as the operational layer; treat the per-regulator detail at the bottom as the reference you go to when something specific needs verification.

Why a Disclosure Workflow Matters Now

Three things changed between 2023 and 2026 that made ad-hoc disclosure unsafe:

• Platforms now enforce. Meta, TikTok, YouTube, and LinkedIn have moved from voluntary AI labels to required ones, and ad-rejection rates for unlabeled synthetic content rose sharply in 2025. Missing the label on a paid placement now triggers review-and-pause cycles that delay campaigns.
• The FTC and EU AI Act create real enforcement risk.The FTC’s revised endorsement guides explicitly call out AI-generated endorsements as a deceptive-practice risk, and the EU AI Act requires clear disclosure on synthetic content that could mislead. Both have teeth: financial penalties, mandated content removals, and reputational damage that outlives the campaign.
• Audiences expect transparency. Consumer trust research consistently shows disclosed AI content slightly outperforms undisclosed equivalents over time because audiences are getting better at spotting synthetic content and reward the brands that level with them. Disclosure is competitive moat, not compliance burden.

Step 1: Classify the Asset’s Synthetic Level

Disclosure obligations scale with how synthetic the asset is. Sort every asset into one of three tiers before publishing:

• Fully synthetic. AI persona + AI scene + AI-composited product. No real human likeness is depicted. This is what most AI UGC platforms produce, including ppl.studio. Disclosure is required on every platform that mandates AI labels.
• Hybrid. Real photo or video with AI-generated background replacement, generative fill, or AI relighting. Real footage of a real person is the load-bearing element; AI is the editing layer. Disclosure is generally required when the AI edit changes what the content represents (e.g. swapping location, weather, or product).
• AI-assisted.Human content with light AI editing — color correction, denoising, AI-cropping. Disclosure is usually not required, but document the AI tooling used in case the standard tightens later.

Step 2: Identify All Destination Platforms

Disclosure rules are platform-specific and per-asset. Before publishing, list every channel the asset will run on:

• Meta (Facebook + Instagram): “AI info” label, applied at upload via the “AI info” toggle. Required on paid and organic. Detection systems may auto-label assets that carry C2PA or SynthID metadata.
• TikTok: “AI-generated content” toggle at upload. Required on assets that depict realistic-looking scenes or people. TikTok’s detection layer increasingly auto-applies the label when undeclared.
• YouTube: “Altered or synthetic content” declaration in the upload flow. Required when AI depicts realistic-looking scenes or persons, especially if the depiction could mislead.
• LinkedIn: “AI-modified” tag, increasingly enforced via metadata detection. Required on synthetic-imagery posts depicting realistic scenes.
• Pinterest, Reddit, Snapchat, X: Each platform’s policy is evolving; check current settings every quarter. As a default, add an on-creative disclosure when the platform lacks a native label.
• Owned web (your site, email, in-app): No platform label exists, so include text disclosure on the asset itself or adjacent caption.
• Paid display, native ad, programmatic: No standard label exists across DSPs — default to on-creative text disclosure for any asset that could be mistaken for real footage.

Step 3: Apply the Platform-Required Label

Each platform has a different name and UI location for the AI-content label. Apply it at upload, not after publication — retroactively labeling an asset that has already started ranking creates a worse user-trust signal than labeling it correctly upfront.

• On Meta, the “AI info” toggle is in the post composer’s advanced settings. For ad creative, the equivalent setting lives in Ads Manager under the creative attributes panel.
• On TikTok, the “AI-generated content” switch is in the post-options screen before publishing. For TikTok Ads, it is part of the asset upload flow in Ads Manager.
• On YouTube, the “altered or synthetic content” question is asked during the upload “Details” step. Answer it accurately.
• On LinkedIn, the AI-modified tag is increasingly auto-applied via metadata. If the platform does not detect the label, add it manually in the post composer.

Step 4: Embed C2PA or SynthID Provenance

Provenance metadata travels with the file across distribution — if a platform’s detection system reads C2PA-compatible signatures or SynthID watermarks, the AI label is applied automatically even if the human uploader forgets.

• C2PA: Cryptographically-signed metadata embedded in the file at generation. Supported by Adobe, Microsoft, OpenAI, Google, Meta, and most enterprise AI image platforms.
• SynthID:Google’s invisible watermarking layer for images, audio, and video. Survives many forms of compression and re-encoding.
• IPTC AI metadata:The IPTC photo-metadata standard now has an “Digital Source Type” field that designates AI-generated, AI-enhanced, or composited content. Embedded in EXIF.

If your generation tool supports any of these, export with provenance enabled. ppl.studio generations carry C2PA-compatible provenance by default.

Step 5: Audit Endorsement and Claim Accuracy

The FTC’s revised endorsement guides specifically address AI-generated content. The rules apply to claims, testimonials, and endorsements that an AI-generated persona makes or appears to make:

• You cannot fabricate a customer review or testimonial and pair it with an AI persona. Quoted reviews must come from real, verifiable customers.
• Performance claims that an AI persona makes (“I lost 20 lbs in 30 days”) must reflect typical results and be substantiated. Same standard as a real-spokesperson testimonial.
• If an AI persona depicts a healthcare professional, expert, or category authority, the credentials being implied must be accurate. A fake AI dermatologist endorsing a skincare brand is a deceptive practice.
• Material connections rules still apply: if your synthetic creative is paid placement, label it as advertising even when the creative itself is synthetic.

The compliant pattern is to use AI personas for the visual layer (a person enjoying the product in a scene) while keeping the quoted text grounded in real customer reviews and verifiable claims. See our blog post on FTC AI content disclosure guidelines for deeper detail.

Step 6: Add On-Creative Disclosure When Ambiguous

Platform-level labels are not always visible. When the asset will distribute in ways that hide the platform UI, add an on-creative text disclosure inside the asset itself:

• Cross-platform reposts, screenshots, and downloads strip platform labels — the disclosure has to live in the pixels.
• Dark posts, paid placements, and ad creative shown in restricted UIs often hide the AI-info label below the fold or in collapsible panels.
• Email, owned-web, and in-product placements have no platform-level disclosure system at all.
• Video B-roll repurposed inside other people’s content loses any platform context.

The standard pattern is a small, legible “AI-generated” or “Created with AI” text in a corner of the visual — not as a disclaimer that hides the content, but as a present, visible note.

Step 7: Document Consent for Any Depicted Person

Brand-owned AI personas — synthetic identities that do not depict any specific real person — do not require a model release. But the moment a real person is involved, full consent documentation applies:

• If you lip-sync a real customer’s voice onto an AI persona, the customer’s consent for endorsement-style use of their voice applies.
• If you blend a real person’s face partially with an AI body, the real person’s right of publicity applies in every state and most countries.
• If you reference-train an AI on a real person’s appearance, even with permission, document the consent scope and term.
• If you use AI to revive a deceased person’s likeness, post-mortem right-of-publicity statutes apply — this is a high-risk area with active state and federal legislation.

ppl.studio’s AI personas are brand-owned synthetic identities — no real-person consent applies because no real person is depicted. The consent rule applies when you blend in real-person content.

Step 8: Apply EU AI Act Labeling for EU Distribution

The EU AI Act requires AI-content disclosure on synthetic media that depicts persons, places, events, or facts in a way that could mislead. For brand creative distributed in the EU:

• Apply the platform-level AI label as in Steps 3–4.
• Add an on-creative disclosure (Step 6) for any asset depicting a person, place, or event in a realistic style.
• Maintain a per-asset record of synthetic level (Step 1) and provenance metadata (Step 4). EU enforcement bodies may request documentation.
• Apply heightened scrutiny to political, news-style, or public-figure content — the EU AI Act treats these categories with stricter disclosure obligations.

UK Online Safety Act and individual member-state laws (especially France, Germany, and Italy) layer additional requirements on top of the EU AI Act. When in doubt, default to the strictest standard among the active jurisdictions.

Step 9: Track the Asset in Your Compliance Log

Maintain a per-asset compliance log. The log is what you need if a regulator, platform, or partner asks — or when policy updates require a retroactive review:

• Asset ID + filename
• Synthetic tier (fully synthetic, hybrid, AI-assisted)
• Generation tool + model used
• Provenance metadata embedded (C2PA, SynthID, IPTC)
• Platforms where the asset is distributed
• Disclosures applied per platform
• Consent records for any depicted real person
• Endorsement and claim substantiation if any persona makes claims
• Generation date + last review date

A simple spreadsheet works for teams under 100 assets per month. Beyond that, integrate the log into your creative asset library so disclosure metadata travels with the asset across reviews, regenerations, and republications.

Step 10: Re-Review When Policy Changes

Platform AI policies are updating multiple times per year. Build a regular review cadence so the team is not caught by surprise:

• Quarterly audit of disclosure settings across active campaigns.
• Same-week response when Meta, TikTok, YouTube, the FTC, or the EU AI office publishes a policy update.
• Subscription to platform policy mailing lists (Meta for Business, TikTok Newsroom, YouTube Creator Insider, LinkedIn Engineering).
• Annual review by legal counsel familiar with AI and advertising law.

The Pre-Publish Compliance Checklist

Print this checklist or pin it in your creative-ops workspace. Use it once per asset before publishing:

Where ppl.studio Fits

The disclosure workflow is easier when the underlying production tool is designed for compliance from the start. ppl.studio produces brand-owned synthetic personas (no model release ever required), embeds C2PA-compatible provenance metadata in every generation, and composites your real product into generated scenes so endorsement-style claims can stay grounded in real customer reviews. The team is responsible for applying platform labels and on-creative disclosures — the production tool stops being the compliance bottleneck.

See our related coverage: FTC AI content disclosure guidelines, C2PA, SynthID, model release, and synthetic media.