YouTube to automatically label realistic AI-generated videos

YouTube has announced it will begin automatically applying labels to realistic AI-generated videos on its platform, including cases where creators do not voluntarily disclose the use of artificial intelligence.

The new system, which uses internal detection signals, takes effect from May 2026. It marks a significant shift from the platform’s earlier policy that relied mainly on creators’ self-disclosure for realistic synthetic content.

According to the announcement, if YouTube’s systems detect “significant photorealistic AI use” in a video, it will apply the label automatically even without creator input. Creators who disagree with the labelling can update their disclosure status in YouTube Studio, though some labels, such as those on content created with YouTube’s own tools like Veo or Dream Screen, or videos carrying C2PA metadata, will be permanent.

Music industry implications

The development has particular relevance for music. Photorealistic AI-generated music videos will carry the prominent label, while stylised or animated videos will not. This creates an incentive for artists and labels experimenting with AI visuals to opt for non-photorealistic formats to avoid the label.

YouTube is also making the AI disclosure labels more visible: for long-form videos, the label will appear directly below the player, while Shorts will show it as an on-screen overlay. Content that is animated, unrealistic, or only slightly altered will continue to have disclosures only in the video description.

The platform emphasised that the labels are intended to improve transparency and do not affect video recommendations or monetisation eligibility.

Background

The move builds on YouTube’s earlier efforts. The company expanded its AI likeness detection tool, which identifies content using creators’ likeness without permission, to all eligible creators aged 18 and above. It had previously been rolled out to selected creators in October 2025 and later to major talent agencies in April 2026.

YouTube still requires creators to manually disclose the use of realistic AI tools, but the new automatic system aims to make enforcement more reliable.

The changes align with comments made in March by YouTube’s Global Head of Music, Lyor Cohen, who said the platform is strengthening guardrails around AI content, including likeness detection, while fighting low-quality AI-generated material. Cohen quoted YouTube CEO Neal Mohan as saying: “AI will remain a tool for expression, not a replacement.”