Prompting real people: likeness rules, refusals and the labels platforms require

By The NeutronField editorsPublished Updated

Short answer

Generator rules and the law both treat real people as the hardest case. OpenAI's terms forbid reproducing anyone's likeness without consent, Google bans impersonation meant to deceive, and platforms such as YouTube and Meta require you to disclose realistic AI content. In the EU, deep fakes must be disclosed from August 2026, and US law now criminalises non-consensual intimate deepfakes.

What generators allow

The image companies write their rules about consent and deception rather than about names. OpenAI's service terms say you may not use its visual features "to reproduce the likeness of any person without express consent and all necessary rights" (OpenAI service terms). Google's generative AI prohibited use policy, which covers the Gemini app and API, bans "impersonating an individual (living or dead) without explicit disclosure, in order to deceive" and anything that facilitates non-consensual intimate imagery (Google prohibited use policy).

frame 62An empty portrait studio with a stool in front of a grey backdrop and a softbox light
▸ 62NF 08
Frame 62. The pictures on this site leave people out on purpose.
We do not run refusal tests for this page: running prompts for real people's likenesses is exactly what the rules restrict, and the images on this site avoid identifiable faces on purpose. What the policies make clear is that a model producing an image does not make the image allowed.

Public figures and private people

A famous person and a private one are not treated the same way in practice, but neither gives you permission by default. Satire and commentary about public figures are common and often lawful; realistic images that could be mistaken for real events are where the rules bite, because the harm is deception.

For private people the bar is simpler: without their consent, do not make realistic images of them. That covers classmates, colleagues and ex-partners as much as anyone, and it is where the strictest laws apply.

Labels platforms require

YouTube requires creators to disclose AI-generated or altered content that seems realistic, including content that "makes a real person appear to say or do something they didn't do", and labels uploads automatically when they contain C2PA metadata (YouTube Help). Meta requires people to disclose photorealistic video or realistic audio that was digitally created or altered, and may add a more prominent label where content risks misleading people on a matter of public importance (Meta community standards); its labels now read "AI info" (Meta transparency).

A signed content credential in the original file helps these labels apply automatically, which is one reason not to strip it. See photorealistic prompts and AI video generators.

What the law says

In the European Union, Article 50 of the AI Act requires providers to mark AI outputs in a machine-readable way, and requires anyone who publishes a deep fake to disclose that it was artificially generated or manipulated; artistic, satirical and fictional works only need a disclosure that does not spoil the work. Article 50 applies from 2 August 2026 (Regulation (EU) 2024/1689).

In the United States, the TAKE IT DOWN Act, signed in May 2025, makes it a crime to knowingly publish non-consensual intimate images, including AI-made "digital forgeries", and since May 2026 requires covered platforms to remove reported images within 48 hours (Public Law 119-12). Other countries have their own rules, and none of this page is legal advice.

For selling images of people, model releases apply; see selling AI images. For edits to real photographs, see edited with AI.

A missing label does not mean an image is human-made.

Sources

  1. OpenAI: Service terms (accessed 2026-11-02)openai.com
  2. Google: Generative AI prohibited use policy, modified 2024-12-17 (accessed 2026-11-02)policies.google.com
  3. Meta: Community standards, misinformation (accessed 2026-11-02)transparency.meta.com
  4. Meta: Labeling AI content (accessed 2026-11-02)transparency.meta.com
  5. YouTube Help: Disclosing use of GenAI content (accessed 2026-11-02)support.google.com
  6. EUR-Lex: Regulation (EU) 2024/1689 (AI Act), Article 50 (accessed 2026-11-02)eur-lex.europa.eu
  7. US Congress: Public Law 119-12 (TAKE IT DOWN Act) (accessed 2026-11-02)congress.gov

Questions

Can I make an AI image of a real person?

Only with care. OpenAI's terms forbid reproducing anyone's likeness without consent, Google bans deceptive impersonation, and many platforms require disclosure of realistic AI images of people.

Do I have to label AI images of real people?

On YouTube and Meta, realistic AI content must be disclosed, and in the EU deep fakes must be disclosed from 2 August 2026 under Article 50 of the AI Act.

Are AI images of celebrities allowed?

Satire and commentary are treated differently from realistic fakes, but no generator's rules give blanket permission. Realistic images that could deceive are where policies and laws apply.

Is it illegal to make AI nude images of someone?

Publishing non-consensual intimate images, including AI-made ones, is a federal crime in the US under the TAKE IT DOWN Act, and Google's and OpenAI's policies forbid making them. Many other countries have similar laws.

Related reading