Evidence reviewed 5 September 2026 by 404 Models Editorial. Original publication date retained.

Direct answer

Our operating recommendation is clear, early disclosure of public synthetic personas and material brand relationships. This is not a claim that every AI-assisted asset has the same legal labelling duty: applicable law, platform rules, realism, interaction, and campaign context determine the requirements. Have counsel review market-specific obligations and regulated categories.

What disclosure has to cover

There are usually two questions. First: is the identity or media synthetic? Second: is there a material brand relationship or paid endorsement? A synthetic model can still be an ad. A disclosed brand account can still need platform AI labels when altered or generated media is used.

Use a disclosure matrix before publishing. Map each content type to profile disclosure, caption disclosure, platform AI label, paid partnership label, landing page note, and internal approval owner. The matrix should be part of the content calendar, not a separate legal file that nobody opens.

Official guidance to check

US: Disclose the Relationship, Not Only the Technology

For US endorsements, the FTC focuses on unexpected connections that could affect credibility, including payments, gifts, and employment. An AI label does not explain who commercially controls the recommendation. The Endorsement Guides interpret Section 5 of the FTC Act; they are not themselves a separate statute or a safe harbor. FTC endorsement guidance, revised in 2023. Put required disclosures with the endorsement, not solely in a profile or behind an expansion control. Video endorsements need disclosure in the video, not just its description. FTC Disclosures 101, November 2019.

A disclosed avatar is not automatically a lawful testimonial. The FTC’s rule FAQ explicitly says virtual influencers are not categorically prohibited, but fake or false underlying testimonials can violate the rule, effective 21 October 2024. Do not present a fictional character as a real customer reporting product experience that never occurred. Disclosure does not cure that false impression. FTC Consumer Reviews and Testimonials Rule FAQ.

EU: Separate Provider Marking From Publisher Disclosure

Article 50 transparency duties apply from 2 August 2026. Commission implementation guidance, published 20 July 2026. Article 50(2) addresses providers’ machine-readable marking of synthetic outputs; Article 50(4) addresses deployers’ disclosure of qualifying deepfake images, audio, and video. Required information must be clear, distinguishable, accessible, and supplied by first exposure at the latest. Directly interactive AI personas can also trigger the separate provider notification duty under Article 50(1), unless interaction with AI is obvious in context. AI Act, Article 50.

Do not assume an invented face is exempt. The Commission’s July 2026 interpretation considers subjects that could plausibly exist and whether the presentation could falsely appear authentic. Assess the persona, message, audience, and context together. Commission Article 50 FAQ. The creative-work provision adjusts disclosure rather than eliminating it, and the human-review exception for certain public-interest text does not exempt human-approved deepfake video. Article 50(4).

The AI Omnibus gives providers of relevant systems placed on the market before 2 August 2026 until 2 December 2026 to meet Article 50(2). This is not a postponement of deployer disclosure under Article 50(4). Regulation (EU) 2026/1744, Article 1(39)(b).

Platform Labels Remain a Separate Check

YouTube requires disclosure of realistic AI-generated or meaningfully altered content, while distinguishing minor edits and non-realistic content. TikTok requires labels for realistic AI-generated images, audio, and video. These labels address synthetic media, not the adequacy of sponsorship disclosure or product substantiation. Check eligibility and labelling for the actual placement before publishing. YouTube guidance, TikTok guidance.

Brand-safe disclosure patterns

Put the persistent identity disclosure in the profile or account bio when the account is a synthetic persona. Use campaign-level disclosure when a post is sponsored, paid, gifted, or connected to a brand relationship. Use platform-native AI labels when the platform provides them and the content qualifies. Keep the wording plain: audiences should understand it quickly.

FAQ

Do AI influencers always need disclosure?

Not every internal render needs public disclosure, but public synthetic personas and synthetic endorsements should be handled transparently. The safer brand posture is to disclose early and avoid ambiguity.

Can disclosure hurt performance?

Viewer responses can differ. YouTube says its AI disclosure does not itself limit audience or monetization eligibility; that is platform policy, not proof of unchanged conversion. Test compliant creative treatments rather than removing a required disclosure. YouTube disclosure guidance.

Is this legal advice?

No. This article is an operating guide for brand teams. Have counsel review regulated categories, claims, talent consent, and market-specific obligations.

Related resources

Related: Virtual influencer IP, Human model question, Contact 404 Models.

Research updates

Research updates reviewed July 20, 2026: EU AI Act Article 50 disclosure guide, C2PA field guide, FTC Disclosures 101, YouTube synthetic-content disclosure.