LEDE
Tavus says 26 of 54 people on a one-minute video call thought its new Griffin model was human. The company’s own test results suggest AI video calls are becoming increasingly difficult to distinguish from real interactions. The Griffin model has not yet been released to retail customers.
KEY FACTS
- Griffin is a new AI model developed by Tavus.
- 26 out of 54 people thought Griffin was human during a one-minute video call.
- Results were reported by Tavus, not an independent source.
- Griffin is not yet available to retail customers.
THE STORY
What is Griffin?
Griffin is an AI model created by Tavus, designed to generate highly realistic video representations of human faces and expressions. According to the company, during a one-minute video call test, 26 out of 54 participants believed the model was a real person. This means roughly 48% of participants were fooled. Tavus has not yet released Griffin to customers, but the internal test results suggest AI video calls are nearing a point where real and synthetic interactions blur.
Why AI Video Calls Matter
The rise of hyper-realistic AI video calls raises serious concerns about misinformation and trust. If models like Griffin can convincingly mimic humans, bad actors might use them for scams or impersonation. While Tavus has not deployed Griffin commercially yet, the technology is advancing quickly and regulatory responses may follow.
How Did We Get Here?
Advances in generative AI and deepfake technology have made synthetic media increasingly lifelike. Tavus’ Griffin joins other high-profile projects aiming to create realistic avatars for video conferencing. Still, the field remains experimental and public access is limited.
WHAT WE KNOW — AND WHAT WE DON’T
Verified by the source:
- Griffin is a new AI model by Tavus.
- 26 of 54 people thought it was human during a test call.
- Tavus reported these results independently.
- Griffin is not yet available to retail customers.
Still unconfirmed:
- Details about how the test was conducted.
- Independent verification of the 26-of-54 result.
- Timeline for public release or broader testing.
WHY IT MATTERS
The rapid improvement of AI video calls has major implications for privacy, security, and trust online. As these systems grow more convincing, users and institutions must prepare for potential misuse.
WHAT TO WATCH
Tavus has not announced a public release date for Griffin. Experts will be watching how regulators respond to advances in AI video call technology.