AI tools like Google’s new product Omni model threats to make reality harder to distinguish from AI fiction, but YouTube is taking an important role in verifying the origin of videos. After AI documentation is published in 2024, Google will start using AI-based video labels, and the site will no longer rely on downloaders to disclose when they use AI tools to create a video.
When YouTube first tried it use AI videos in 2024, it was almost worthless. AI animations at the time almost always looked weird or inconsistent. In just a few years, AI models like Seedance, Runway, and Google’s Veo have improved the accuracy and consistency in AI video—Spaghetti and more accurate than before.
Realizing this, YouTube is making AI content popular and making itself part of the process. Producers are still required to indicate when uploading videos that they were created with the help of AI tools. However, downloaders had no incentive to tell the truth about this in the past. Starting this month, YouTube will use “new internal signals” to advertise AI. This seems to apply to videos that show “heavy use of AI graphics.”
Simplified AI Labels & Auto-Detection: What You Need to Know
Google doesn’t know which signals will be found in its AI detection system – we’ve asked for more information and will update if we hear anything. The blog post mentions two ironclad initiatives: C2PA metadata to reveal the source of AI and the use of popular Google tools like Veo. Producers who believe their videos have been incorrectly labeled as AI may be interested, but not if the site shows the promotion as AI for either reason. The records are “eternal.”
