Policy and regulation illustration
YouTube said on 27 May 2026 that it is making AI-generated video easier to spot, both by moving disclosure labels into more visible places and by applying them automatically when creators do not. The company has required creators to disclose realistic AI use since 2024. The new post, from the YouTube Team, is about making that disclosure harder to miss and less dependent on voluntary honesty. For long-form videos, YouTube said the label for photorealistic or meaningfully AI-altered content will sit directly below the player, above the description. On Shorts, it will appear as an overlay on the video itself. That format, the company wrote, is now the single label for all photorealistic and meaningfully AI-altered or generated content. Unrealistic, animated or only slightly altered material will still carry disclosure in the expanded description rather than on the main stage. The more consequential change is automatic detection. YouTube still requires creators to say when they used realistic AI, but starting in May 2026 it is rolling out internal signals to identify AI-generated content. If a creator does not specify whether they used AI, and YouTube’s systems detect significant photorealistic AI use, the platform said it will apply a label anyway. Creators who think the detection is wrong can update the disclosure status in YouTube Studio. YouTube said disclosures will remain permanent in a handful of cases, including content made with YouTube’s own tools such as Veo or Dream Screen, and content that carries C2PA metadata showing it was fully generative AI. YouTube emphasised that a disclosure label alone does not change whether a video is recommended or whether it can earn money. That is an important limit. A labelled deepfake of a politician, a celebrity or a local business owner can still spread; the label is context, not a takedown. Detection, the company implied, will keep improving, which also means false positives and false negatives are likely during the rollout. YouTube did not publish accuracy rates. For viewers in places where AI impersonation is already used in investment and romance scams, a label is useful only if people look for it and then refuse to act on a video that asks for money, logins or personal data. For creators, the Studio appeal path is the place to correct an over-eager label, not a public fight in the comments. What to do • Watch for the new on-player and Shorts overlay labels before sharing a video that appears to show a real person saying something sensational. • Do not treat a missing label as proof that footage is authentic. YouTube said detection is still rolling out. • Creators should disclose photorealistic AI at upload time and, if a label looks wrong, change the status in YouTube Studio rather than stripping watermarks. • If a video of you or your organisation is being used to sell a product or request money, use YouTube’s privacy complaint process and warn customers on a channel you control. • Small organisations should tell staff that “the CEO said so on video” is not verification; confirm payment and identity changes out of band. Primary source: YouTube Blog, “Improving AI labels for viewers and creators”, 27 May 2026.

Sources

  1. YouTube Blog — Improving AI labels for viewers and creators2026-09-24

Cyber71 links the primary material behind each report so readers can check it directly. External pages are not under our control and may change.

All news

Need help with a security issue?

Describe the situation and we will open a private case for follow-up.

Get Security Help