AI image disclosure — rules, labels and examples

Direct answer

AI images need disclosure when they are photorealistic and generated or manipulated in ways that could falsely appear authentic — synthetic models, altered appearances, fake events or realistic generated scenes. EU AI Act Article 50(4) requires deployer disclosure for such deepfakes from 2 August 2026. Instagram, TikTok and YouTube require native AI labels for realistic AI imagery. Partial use, like generative fill around a real product, needs wording that discloses exactly the AI part.

Working with an edited image?

Inspect its embedded AI signals before posting. This does not predict whether Instagram or another platform will definitely add an AI label.

Inspect image signals

Applies when

Photorealistic fully generated images.
Generated or altered people, faces, bodies or models.
Generated scenes or events that look filmed or photographed.
Generative fill that extends reality convincingly.

Doesn't apply when

Clearly stylised, illustrated or unrealistic imagery.
Minor retouching that does not change what the image appears to show.

Worked decision

Example

Real product, AI-generated beach background

Disclosure recommended

The product is real; disclose the generated background specifically to stay truthful and platform-safe.

Next steps

Required action

Use the native AI label on the publishing platform.
Word the disclosure to match the actual AI component.
Record tool, prompt scope and approver in your AI Usage Manifest.

Publishing layer

Platform differences

Law, platform and client layers

EU AI Act Article 50(4) requires deployers to disclose realistic AI-generated or manipulated image, audio and video resembling real persons, objects, places or events. Use the platform’s native label and keep your own evidence record.

Evidence

Source

Regulation (EU) 2024/1689 (Artificial Intelligence Act)European Union · effective from 2024-08-01 · last checked 2026-09-12
Labeling AI Content — Meta Transparency CenterMeta · last checked 2026-09-10
About AI-generated content — TikTok SupportTikTok · effective from 2024-05-01 · last checked 2026-09-23

Freshness and accountability

Last verified

Applies from 2026-08-02 · Last reviewed 2026-09-10 · Current

Run this exact scenario

Relevant content, AI-use and platform inputs are preselected. Confirm the remaining facts before relying on the result.

Open prefilled checker

Who is affected

Performance agencies generating ad creative at scale.
E-commerce teams using virtual models or generated backgrounds.
Social teams compositing real products with AI scenes.

More scenarios

Scenario	Outcome	Why
Fully generated photoreal model wearing real clothes	Disclosure required	Realistic synthetic person in commercial content — native label plus wording.
AI upscaling of a real photo	No disclosure identified	Enhancement does not alter what is depicted; record it in the manifest.

Practical marketing examples

“Background generated with AI. Product photographed.”
“Model is AI-generated. Clothing is real.”

Disclara separates binding law, platform policy, client policy, technical provenance and recommended wording. Information on this page is not legal advice. See editorial policy.