Do I have to label AI-made content?

You're scrolling and notice a small tag under a video. It says the post was made with artificial intelligence (AI), or it offers "AI info." Who put it there? Sometimes the person who posted it. Sometimes the app, on its own. And in some places, a law asks for it. As of October 2026, whether you have to add a label depends on what you made, where you post it, and where the people seeing it live.

Labels come in two kinds.

• One you can read: a caption, a tag, or a line of text saying AI made or changed the content.
• One hidden in the file: machine-readable data, like a watermark or content credentials, that software can check.

Laws and apps can ask for either one, or both.

The broadest rule so far comes from the European Union (EU). Its AI law, the AI Act, has a section on transparency called Article 50. It asks the companies that make AI tools to mark what those tools produce in a machine-readable way, so the output can be detected as AI-made. And it asks the people and businesses who publish certain content to say so plainly. These rules have applied since August 2, 2026.¹²³

Under the EU rule, deepfakes get a visible label. Those are realistic images, audio, or video of people, places, or events that were made or changed by AI, and whoever publishes one has to disclose it. Text gets a label in one case: when AI wrote it and it's published to inform the public on matters of public interest, like a news-style article. Even then, the label isn't needed if a person reviewed the text and someone takes editorial responsibility for it.¹²

The same article covers conversations. When you talk to an AI system, like a customer-service chatbot, you should be told you're talking to a machine, unless that's already obvious.¹

Outside Europe, the rules are a patchwork. In the United States (US), some states have passed their own. New York, for one, now asks for a clear label on ads that use an AI-made performer. Apps add rules too. As of October 2026, YouTube asks creators to disclose realistic content that was meaningfully altered or made synthetically, such as a video of an event that never happened. TikTok asks creators to label realistic AI-generated content, and it can add labels itself by reading content credentials. Policies like these change often, so the app's own help page is the place to check.⁴⁵⁶

If you're the one watching, a label is useful information: it tells you a machine had a hand in what you're seeing. But a missing label proves nothing. People skip them, and hidden data can be stripped when a file is saved again. So when a surprising video has no tag, you still don't know how it was made.

If you make content for a business, the rules that apply depend on what you made and where it will run, and when the stakes are real, that's a question for a lawyer. For everyday posts there's a simple habit to try. Before you share something made with AI, open the app's upload screen and look for its setting for AI or altered content.