YouTube announced automatic detection of photorealistic AI-generated or altered videos beginning May 2026, with labels placed below the player for long-form content and as overlays for Shorts. The platform will apply labels when internal signals detect significant AI use if creators omit disclosure, while retaining manual override except for Veo, Dream Screen, or C2PA-marked files. Labels do not affect recommendations or monetization. YouTube's rollout follows its 2024 manual disclosure requirement and incorporates C2PA metadata standards referenced in the announcement. Comparable policies appear in Meta's April 2024 AI labeling update and OpenAI's content credential guidelines, both requiring provenance signals on synthetic media. Automatic enforcement at upload time marks the first platform-scale application of detection signals without creator input. The policy ties disclosure permanence to specific tool outputs and metadata flags, creating a verifiable chain for photorealistic content while leaving non-photorealistic alterations in expanded descriptions only.