If you publish photorealistic AI images of a fictional person, say so where people see them: the platform's AI label on every post and a line in the bio. In the EU this has been a legal duty since 2 August 2026 for anyone using AI images in a professional activity, and the European Commission's draft guidelines say a realistic persona who could plausibly exist is covered. This guide is not legal advice.
What the EU AI Act requires
Article 50(4) of the AI Act says deployers of AI that generates a "deep fake" must disclose that the content is artificially generated or manipulated. A deep fake is content that resembles existing persons, objects or places and would falsely appear authentic. A deployer is anyone using an AI system, except in a purely personal, non-professional activity, so a hobby account may fall outside, while a brand persona or a monetised influencer is inside.
- Fictional people count. The Commission's guidelines on Article 50 (approved as a draft in July 2026, formal adoption pending) say it is enough for a subject to resemble someone who can plausibly exist, and include realistic AI-generated avatars and personas.
- "It's fiction" only goes so far. Evidently artistic, creative or fictional work gets a lighter duty: a disclosure that doesn't spoil the work. The guidelines read that strictly and exclude purely commercial content; their own example of what doesn't qualify is a realistic synthetic influencer trying out a sponsored product.
- The label must be visible. Invisible watermarks added by the tool maker are the provider's separate duty under Article 50(2); they don't disclose anything to your audience.
- Fines for breaking Article 50 go up to €15 million or 3% of worldwide annual turnover.
Did the Digital Omnibus delay it?
No, not for you. The Digital Omnibus regulation (EU 2026/1744) gives providers of generative AI placed on the market before 2 August 2026 until 2 December 2026 for machine-readable marking. Your duty to disclose deep fakes didn't move. The Commission's Code of Practice on marking and labelling is voluntary.
What the platforms ask
- TikTok: its community guidelines require you to label AI-generated content that shows realistic-looking scenes or people; unlabeled content may be removed, restricted or labeled by TikTok.
- YouTube requires disclosure of realistic AI-generated or altered content: in YouTube Studio, answer "Yes" under AI use. Repeated failure can lead to removal or suspension from the Partner Program.
- Instagram and Facebook: Meta requires its AI disclosure for photorealistic video and realistic audio. Still images aren't named in that rule, but the "AI info" label is there; use it.
- LinkedIn shows a Content Credentials icon when a file carries them and, in some countries, lets you self-attest AI use. It's optional; we'd still say it in the post.
When the persona sells something
Advertising rules stack on top. In the US, the FTC's endorsement guides say that if an ad shows people who appear to be actual consumers, they must be real consumers or the ad must disclose that they aren't. In the EU the usual hidden-advertising ban applies, and commercial persona content gets no lighter AI Act treatment. Label it as an ad and as AI. Never use a persona for reviews or testimonials; Dittoshot's terms forbid fake reviews outright.
A checklist for every AI persona
- A plain line in the bio: "AI-generated character. Not a real person."
- The platform's AI label switched on for every post, not only the first.
- No captions or replies that claim the persona is a real human.
- Sponsored posts carry both the ad disclosure and the AI disclosure.
- Keep the disclosure readable: not buried after thirty hashtags.
Dittoshot makes images only and never posts for you, so the label is yours to add. See how Dittoshot works.