YouTube now requires to label their realistic-looking videos made using AI

Original Article ↗ Hacker News Discussion ↗

Scope of YouTube’s Policy

  • Many note the policy covers “altered or synthetic” realistic content broadly (VFX, animation, AI), not just AI.
  • Key triggers: making real people appear to say/do things they didn’t, altering footage of real events/places, or generating realistic scenes that never happened.
  • Non‑required disclosures include fantasy scenes, green screens, filters, AI for scripts/thumbnails, captions, and upscaling.

Regulation Drivers and Motives

  • Several comments link this directly to compliance with the EU AI Act and similar emerging rules elsewhere.
  • Others suggest Google is also protecting training data quality and collecting labeled data to train AI‑detection models.
  • Some see it as culture‑shaping or a preemptive move to fend off stricter regulation.

Enforcement and Practical Challenges

  • Heavy skepticism about self‑reporting: good actors will label; bad actors, scammers, and propagandists will not.
  • Concern that YouTube’s automated moderation and appeals processes will mislabel or wrongly penalize creators.
  • Gray areas: beauty filters vs face swaps, VFX vs AI, AI‑assisted vs AI‑generated music, satire vs deception, voice cloning of oneself vs others.
  • Worry that labels will become ubiquitous and meaningless, like Prop 65 warnings or cookie banners.

Trust, Provenance, and Cryptographic Solutions

  • Recurrent theme: realistic video can no longer be assumed true; society must shift to skepticism by default.
  • Some advocate cryptographic signing and provenance trails for cameras and media, but others argue these are spoofable, hard to deploy at scale, and easily undermined.
  • There is debate over whether such mechanisms address authenticity (who made it) but not accuracy (whether it’s true).

Power, Censorship, and Cultural Impact

  • Critics fear the labels will legitimize YouTube deciding what’s “real,” enabling selective takedowns or censorship of inconvenient real footage by calling it “AI.”
  • Others welcome any step that increases transparency and lets viewers filter or deprioritize synthetic content.
  • Broader anxiety about a “post‑truth” environment, but some argue this will force healthier skepticism and a focus on trusted sources and context.