cd /news/artificial-intelligence/new-ai-could-identify-the-ai-behind-… · home topics artificial-intelligence article
[ARTICLE · art-78786] src=qazinform.com ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

New AI could identify the AI behind fake videos

Researchers from the University of California, Riverside, YouTube, and Google DeepMind have developed SAGA (Source Attribution of Generative AI Videos), an AI system that can identify whether a video was AI-generated and determine which specific AI model created it, using only 0.5% of labeled training data. The framework, tested on 19 different AI models, outperforms existing detection methods and could strengthen digital forensics as AI-generated video becomes more widespread.

read2 min views4 publishedJul 29, 2026
New AI could identify the AI behind fake videos
Image: Qazinform (auto-discovered)

Researchers from the University of California, Riverside, YouTube, and Google DeepMind have developed a new artificial intelligence system that can identify not only whether a video was generated by AI, but also determine which AI model created it, according to a newly published research paper, Qazinform News Agency correspondent reports.

The system, called SAGA (Source Attribution of Generative AI Videos), can attribute it to a specific generator, development team, model version, or generation method, providing more detailed information for digital investigations and content verification.

According to the researchers, the framework works across five levels of identification: distinguishing real from AI-generated videos, identifying whether a video was created from text or an image prompt, recognizing the underlying model version, identifying the development team, and pinpointing the exact AI generator.

A key feature of SAGA is its ability to learn from very limited labeled data. The researchers say the system achieved results comparable to fully supervised methods while using only 0.5% of the source-labeled training data for each AI model.

The researchers also introduced what they call Temporal Attention Signatures, visual patterns that reveal subtle differences in how AI video generators create motion from frame to frame. These signatures provide an explanation for why different AI models can be distinguished and may also help identify videos produced by previously unseen generators.

To evaluate the system, the team tested SAGA on public datasets containing videos generated by 19 different AI models. The study reports that the framework consistently outperformed existing video detection methods and remained effective even when evaluating videos created by generators it had not encountered during training.

Researchers say SAGA could strengthen digital forensics and improve transparency as AI-generated video becomes more widespread.

Earlier, Qazinform News Agency reported that scientists invented fake illness and AI spread it as real.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @university of california, riverside 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/new-ai-could-identif…] indexed:0 read:2min 2026-07-29 ·