The Change, Plainly Stated

Starting this month, YouTube is doing two things it should have done a long time ago.

First, AI disclosure labels are moving to a place humans can actually see them. For regular long-form videos, the label — displaying "AI" next to an information icon — will now appear directly below the video player, above the description, according to TechCrunch. Previously, you had to click "expand description" and look under "How this content was made." Nobody was doing that.

For YouTube Shorts, the same AI label will appear as an on-screen overlay. The Verge reported YouTube had already been quietly testing a variation of this overlay.

Second, YouTube is deploying "new internal signals" to automatically detect and label videos that make significant photorealistic AI use — even if the creator never disclosed anything. Variety confirmed this rollout begins in May 2026.

Two Years of Hiding the Ball

YouTube introduced AI disclosure requirements over two years ago, per TechCrunch. Creators were required to flag realistic AI content in Creator Studio. The label existed. It just lived somewhere almost no viewer would ever find it.

The platform buried disclosures in expanded descriptions for most content. The only exception: sensitive topics like health or news got more prominent warnings. Everything else? Hidden. That was liability cover dressed up as transparency.

The Verge's Jess Weatherbed said YouTube's labeling practices have been "inconsistent" until now. The system was broken by design.

What Triggers a Label

According to YouTube's own support documentation via Google, creators must disclose AI use when content makes a real person appear to say or do something they didn't do, alters footage of a real event or place, or generates a realistic scene that never happened.

Things that don't require disclosure: animated content, beauty filters, color adjustments, AI-assisted scripts, caption generation, video upscaling, and gameplay footage. The line is photorealism and potential deception — not AI use in general.

YouTube confirmed to Variety that its underlying policy hasn't changed. What changed is enforcement: the platform will now apply labels automatically when its systems flag significant photorealistic AI use, whether or not the creator said a word about it.

When Labels Stick Permanently

Creators who think they've been wrongly flagged can dispute it through YouTube Studio. But according to TechCrunch, two categories of AI labels cannot be removed:

• Content made with YouTube's own AI tools, specifically Veo or Dream Screen
• Content carrying C2PA metadata that identifies it as fully AI-generated

C2PA is a technical standard from the Coalition for Content Provenance and Authenticity. OpenAI recently committed to it, joining Nvidia, Kakao, and Eleven Labs. When a video carries that metadata, YouTube treats it as definitive — end of discussion.

The Google I/O Connection

TechCrunch made a connection the other outlets danced around. This announcement came one week after Google unveiled new Gemini AI capabilities and video generation tools at its I/O developer conference — advances that enable high-quality, physically realistic video generation.

Google just released more powerful AI video tools and simultaneously announced YouTube will get better at labeling AI video. The company is selling the problem and the warning label at the same time. Google is at least doing something about it, even if the timing raises questions.

Deepfake Detection Is Also Expanding

This labeling update is running alongside a separate expansion of YouTube's AI deepfake detection tool, which according to TechCrunch now allows any adult to scan YouTube for face matches. It started with celebrities, public figures, and politicians. Now it's open to everyone.

This is a meaningful escalation in protecting regular people — not just famous ones — from having their likeness used without consent.

What Mainstream Coverage Is Missing

Most tech outlets are treating this as a routine platform update. It isn't.

YouTube has over 2 billion logged-in users monthly. The volume of AI-generated video flooding the platform is accelerating — not stabilizing. The label system that existed for two years was, functionally, useless for the average viewer. This fix is overdue, and the only reason it's happening now is that AI video quality has gotten good enough to be genuinely indistinguishable from real footage.

Why did it take this long? YouTube had the data. It had the disclosure system. It chose to keep labels where viewers couldn't see them.

What This Means for Regular People

If you watch YouTube — and roughly 2 billion of you do — AI-generated video pretending to be real footage just got a visible warning label.

If you're a creator using AI tools honestly, nothing changes. Disclose it, the label appears, viewers know. If you're a creator trying to pass AI content off as real, YouTube's systems will eventually catch and label it anyway.

The platform is late. But the direction is right. Keep the pressure on to make sure they actually follow through this time.