AI-content disclosure & labeling rules by platform

YouTube confidence: high — verified

YouTube requires creators to disclose when realistic content is made with AI: use the "AI use" disclosure setting when a real-looking person, place, or event was generated or materially altered synthetically. The obligation is triggered by realism — clearly fantastical or animated content (your neon noodle detective in a stylized world) is exempt from the label. Disclosure never appears as a strike; skipping it on realistic content can. Source: support.google.com/youtube/answer/14328491 (verified during this product's build).

TikTok confidence: medium — verify

TikTok's community guidelines require labeling realistic AI-generated content with its AI-generated content (AIC) tag, and AI content depicting realistic scenes carries restrictions regardless of label. TikTok pages were not reachable from our build environment, so SerialStorm seeds these rows at medium confidence with an explicit verification note — check TikTok's community guidelines before relying on specifics. Branded-content disclosure (paid promotion) is a separate toggle.

Instagram / Facebook confidence: medium — verify

Meta requires disclosure of photorealistic AI-generated video in several contexts and applies its own AI info labels; branded-content/paid-partnership tagging is separate. As with TikTok, Meta's policy pages were not reachable during our build — medium confidence, verify against Meta's AI-generated content policy.

The operator's rule of thumb

• Realistic → label everywhere, even where it's arguably exempt. A missed label costs more than an extra one.
• Stylized/fantastical → usually exempt on YouTube, still worth an honest caption note.
• Paid promotion of any kind → the platform's paid-promotion/branded-content toggle, always.
• Keep evidence: SerialStorm's compliance checks store the rule's as-of date and confidence next to each result, so you can show why you decided what you decided.

Run a check per episode and platform in the workbench before scheduling.