New 2026 rules will require labels on AI-made text, images, and video

In short: Governments and major platforms are moving toward mandatory labels for AI-generated content in 2026, so people can tell when something was made by a machine.

What's going on

More countries are requiring clear labels when AI is used to create content that looks or sounds like a person made it. This includes text, images, audio, and video. The goal is simple, make it easier to trust what you see online, especially when people often comment, “This looks like AI.”

South Korea plans to make its AI Content Labeling System fully mandatory starting in January 2026. The rules focus on generative AI services (tools that can produce new text, pictures, or voices from a prompt), deepfakes (fake videos or audio that copy a real person), and commercial advertising. Labels may appear as subtitles, watermarks, or other visible notices. South Korea also includes an exception for some artistic or creative works if a label would hurt the viewing experience.

In the European Union, the AI Act is set to require labeling starting August 2, 2026 for content that is largely made by AI and could be mistaken for human-made. The EU rules also call for “machine-readable” markings, which are hidden details computers can read, like information tucked into a file label (similar to a barcode on a product). There are exceptions when a human has clearly reviewed and edited the content, especially for public interest uses.

Platforms are also tightening their own rules. TikTok, Meta, and YouTube already require disclosures for some AI-generated content, and Google Ads plans “AI Generated” labels on ads starting March 5, 2026.

What to watch

These rules focus on labeling AI-made content, not proving something is human-made. A universal “human-made” logo does not exist yet, so the next debate may be whether people also want a standard label for work created without AI.

Source: The Verge AI