Ars Technica: YouTube's AI labels miss significant AI creations
YouTube requires creators to label AI-generated or significantly altered photorealistic content. However, the policy includes ambiguities that allow many AI creations to go unlabeled.
Ars Technica has highlighted shortcomings in YouTube's policy concerning the labeling of AI-generated content. The platform mandates that creators disclose when AI is used to meaningfully alter or generate photorealistic material.
The policy, however, contains inconsistencies. It applies to "AI-generated music," which may not be photorealistic, but not to "riding a unicorn through a fantastical world," even if such imagery could be photorealistic, albeit implausible.
YouTube's own summary states, "Realistic AI content and meaningful changes require disclosure, while non-realistic or minor edits don’t." This leaves a wide range of AI applications unaddressed by the disclosure requirement.
Unlabeled uses can include everything from "idea generation" to "production assistance," such as using AI tools to create or refine video outlines, scripts, thumbnails, titles, or infographics. Creators are also permitted to clone their own voices for voiceovers or use AI-generated animation of a missile in a fully animated video. Ars Technica suggests these loopholes could lead to significant AI-driven content not being properly identified.