UC Riverside Tool Traces Deepfake Videos Back to the AI That Made Them

Key points

• UC Riverside researchers have released a tool called SAGA (Source Attribution of Generative AI) that identifies which AI model produced a fake video.
• The tool works by detecting "temporal signatures," tiny patterns in how video frames change over time that each AI model leaves behind like a fingerprint.
• A PhD researcher on the project told Dark Reading he cannot reliably tell real videos from AI-generated fakes with the naked eye.
• The team plans to move from detection toward prevention, aiming to stop harmful AI video content before it is ever created.

What is this tool, and why does it matter?

Researchers at the University of California Riverside have built software that can trace a fake video back to the specific AI system that generated it. That matters because simply knowing a video is fake is no longer enough to stop the harm it causes.

The tool is called SAGA, which stands for Source Attribution of Generative AI. It does four things: confirms whether a video is AI-made, names the specific model that made it, identifies that model's version, and records which development team built it.

Deepfakes, meaning videos in which a person's face or voice has been digitally replaced or manufactured from scratch using artificial intelligence, have reached a quality where even specialists struggle to spot them. Rohit Kundu, a doctoral candidate at UC Riverside who works on AI video research, told Dark Reading he cannot reliably tell real footage from fakes himself.

How does SAGA actually work?

Every AI video model leaves behind a kind of invisible fingerprint. SAGA finds it.

The team discovered that different AI generators produce what they call temporal signatures, or T-Sigs. These are subtle, consistent patterns in how one video frame transitions to the next. Give two separate AI models the same instruction, and they will produce similar-looking clips. But the tiny inconsistencies in their frame-to-frame movement differ, and that difference is enough to identify the source.

The researchers built SAGA on top of a foundation model, meaning a large general-purpose AI system used as a base, rather than training it on a narrow dataset. That design choice means the tool is less likely to fail when it encounters video types it was not specifically trained on, which is a common weakness in detection software.

Who is this aimed at, and what happens next?

The immediate audience is AI companies themselves. Kundu's hope is that if a particular model keeps appearing as the source of harmful deepfakes, whether that means fake job interview candidates, fabricated celebrity endorsements, or fraud, that company can be notified and pushed to tighten its content filters.

"If some big group's model is being used, you'd want to let them know: 'A lot of videos were generated using your technology, so you want to put more restrictions,'" he said.

The researchers have published a white paper on SAGA's methodology. Their roadmap now points toward proactive prevention: blocking harmful content at the point of generation, before a fake video ever reaches a social media feed or a hiring manager's screen.

What should ordinary people watch for?

Anyone conducting video interviews, reviewing video testimonials, or watching video news clips online is already operating in an environment where fakes are hard to spot with the naked eye.

Practical steps worth taking now:

• For hiring teams: verify a candidate's identity through a second channel, such as a phone call to a number independently looked up, not one the candidate provides.
• For individuals: treat surprising or emotionally charged video clips with scepticism before sharing them. Check whether a reputable news outlet has verified the footage.
• For businesses: ask your video-conferencing platform or HR software provider what steps they take to detect AI-manipulated video streams.