Policy · 27 May 2026 · 2 min read

YouTube Will Now Automatically Label AI Videos

YouTube is now automatically labelling AI-generated video, forcing builders to treat content provenance and transparency as a core product feature.

Pen-and-ink illustration: a transparent casing being lowered over a. For the story "YouTube Will Now Automatically Label AI Videos".
— Pen-and-ink illustration: a transparent casing being lowered over a. For the story "YouTube Will Now Automatically Label AI Videos". —

What happened

YouTube will now automatically label videos that use significant photorealistic AI. TechCrunch AI reports the company's internal systems will apply these labels, even if creators neglect to disclose their use of AI tools. The change builds on a two-year-old policy that required creators to self-disclose.

The new labels will be more prominent, appearing directly below the video player. Labels are permanent for content made with YouTube's own AI tools or containing C2PA metadata. YouTube says the labels will not impact a video's recommendations or its ability to be monetised.

How the room's reading it

The consensus among platform watchers is that this was inevitable. With models like Google's own Gemini Omni now capable of high-quality video output, self-disclosure was never going to be enough. The move is seen as YouTube getting ahead of the next wave of photorealistic content — a shift from creator responsibility to platform-enforced transparency.

Builders shipping generative video tools are taking note. The permanent labelling tied to C2PA metadata is a strong signal for the entire ecosystem. The chatter on X and developer forums suggests this move forces the whole stack to take provenance seriously. It's no longer a feature you can add later; it's becoming table stakes for getting distribution on the world's largest video platform.

Sailfish's take

We see this as a clear signal for anyone building in the generative media space. The game is no longer about creating undetectable fakes. The platforms are building the guardrails, and fighting them is a waste of energy. We think the interesting work will embrace this — making the synthetic nature of the media part of the point.

We've shipped enough media products to know that user trust is everything. This move gives builders an opportunity to build that trust from the ground up. Instead of chasing photorealism that might get flagged, we'd focus on tools for new creative formats that are proudly synthetic. If you're building a generative AI video tool, this is the week to prioritise your content provenance and watermarking strategy. It's not a sideshow anymore.

Our take — your read?

Be the first to weigh in.

Sources
— END OF DISPATCH — Policy