Do You Have to Disclose AI Content? The Honest Creator’s Guide

The news

What changed this week: invisible labels on AI text

OpenAI announced on October 5 that it will add a hidden watermark, called textGrain, to text that ChatGPT and Codex generate in the European Union. It rolls out over the coming weeks to eligible users on all plans. Developers anywhere can switch it on through the API, but it is off by default and is not turned on for ChatGPT outside the EU.

The watermark works by gently shaping the model’s word choices. You can’t see it, and it survives copy and paste. A detector with a secret key can find the pattern. For now, only approved researchers and expert organizations can use the detector.

It is far from airtight. The Register reported that swapping 10% of words for synonyms cut detection from 92% to 66% in OpenAI’s own tests, and swapping 25% left it detectable only 17% of the time. OpenAI also says a missing watermark doesn’t prove a person wrote something.

The reason for all this is the EU AI Act. Its transparency rules, known as Article 50, took effect on August 2, 2026. Businesses and professionals that publish deepfakes must disclose them clearly, with a lighter touch for obviously artistic, satirical or fictional work. A hidden machine-readable mark alone is not enough.

The honest answer

Do you have to disclose AI content? Here’s the line

Across the major platforms and the new EU rules, one idea repeats: disclose when AI makes something look or sound real that isn’t. Using AI as a helper behind the scenes is a different thing.

Use this quick checklist before you post. If you answer yes to any line, label it:

• Does a real person appear to say or do something they didn’t?
• Did AI change footage of a real event or place?
• Does it show a realistic scene that never happened?
• Is the face, body or voice on screen AI-generated and realistic, including an AI twin of you?
• Is it an ad or sponsored post that uses AI people, reviews or results?

That last line matters beyond platform rules. The FTC’s rule banning fake reviews and testimonials, announced in August 2024, names AI-generated fake reviews as an example of what is prohibited. A label doesn’t fix a fake. Simply don’t make one.

Platform by platform

AI disclosure rules on YouTube, TikTok and Instagram

YouTube. The altered or synthetic content policy asks you to disclose realistic content that is meaningfully altered or generated, including AI music that is the main focus of a video. Beauty filters, lighting fixes, unrealistic fantasy scenes and gameplay don’t count. YouTube may also add labels automatically when content carries Content Credentials or its systems detect AI. Disclosing doesn’t limit your reach or monetization, but creators who keep skipping it can face removed videos or suspension from the Partner Program.

TikTok. Its Community Guidelines require clear labeling when AI or editing realistically depicts people or scenes. In a July 2026 update, TikTok said it has labeled more than 3 billion videos as AI-generated, using creator labels, Content Credentials and invisible watermarks together.

Instagram and Facebook. Meta said in 2024 that people must disclose when they post photorealistic video or realistic-sounding audio that was digitally created or altered, and that accounts that don’t may face penalties.

Your digital likeness

Do you have to disclose an AI twin, avatar or voice clone?

An AI twin is a digital extension of you. It is a creative tool, not a trick, and it stays that way when your audience knows. (New to the idea? Start with what an AI twin is and how creators use one.)

There is one interesting detail in YouTube’s policy: cloning your own voice for voiceovers or dubs doesn’t need a label. A realistic AI video of your face is a different story. Our rule at Pixel Pinkscape is simpler than any platform’s: if it looks or sounds like a real person, tell people it’s AI, and only use a likeness with that person’s consent. Labels are cheap. Your audience’s trust is not.

If you want help creating a consistent, clearly labeled AI twin for your brand, Pixel Pinkscape creative and design work and Pixel Pinkscape Avatar Studios are built for exactly this.

Your takeaway

How to write an AI disclosure policy for your content

Rules tell you the minimum. A short, public AI policy tells your audience the whole truth, once. In August, Tubefilter reported that YouTuber Hank Green published five rules for his own AI use after viewers questioned one of his scripts. You don’t need to copy his rules. You only need your own.

1. List your workflow. Ideas, research, scripts, filming, voice, visuals, editing, captions, thumbnails.
2. Mark each step. Human, AI-assisted with your review, or AI-generated.
3. Draw your red lines. For example: AI never invents facts, reviews or results.
4. Choose your label words. One short phrase you will use every time, like “AI twin” or “AI-generated visuals.”
5. Publish it. Put it in your bio link, channel description or website, and update it when your tools change.

I create [type of content] for [audience] on [platforms]. Here is how I make a typical post: [describe each step and which tools you use].

Write a short, friendly AI use policy (under 150 words) for my audience that:
- says plainly which parts are made by me and which use AI,
- lists anything I will never use AI for,
- gives the exact label I will add when a face, voice or scene is AI-generated.
Then list any posts in my workflow that platform rules on YouTube, TikTok or Instagram would likely require me to label.
Don't add claims I didn't give you. Ask me questions if anything is unclear.

Free resources

Free resources for honest AI content

• YouTube Help: Disclosing altered or synthetic content — the free official guide, with clear examples of what needs a label and what doesn’t.
• Elements of AI — a free online introduction to AI for non-experts from MinnaLearn and the University of Helsinki, useful for understanding how these tools work.
• The Pixel Pinkscape Dreamers community — a free Facebook group where creators compare how they label and talk about AI work.
• How to tell if a video call is AI — our earlier guide to the other side of disclosure: protecting yourself when someone else doesn’t label.

FAQ

Frequently asked questions