Tavus unveils Griffin AI model, with nearly half of test subjects mistaking its video avatar for human
Tavus, an AI company, has unveiled its new Griffin AI model, which generates lifelike video avatars capable of participating in one-minute video calls. Tavus claims that in internal tests, nearly half of the participants (specifically 48%) mistook the AI avatar for a real person. This result is interpreted by some outlets as the AI passing a "Video Turing Test." While multiple outlets report on Tavus's claims regarding the AI's ability to mimic human interaction on video calls, the central claim about the 48% figure and its interpretation as a "Video Turing Test" originates from Tavus itself, likely through a preview or company announcement.
7 articles from 6 outlets covered this story. Their coverage differs on 2 points. The underlying claim is sourced from a demo.
What do all outlets agree on?
6 outlets covered “Tavus unveils Griffin AI model, with nearly half of test subjects mistaking its video…”. All of them report the following:
- Tavus developed an AI model named Griffin
- The AI generates video avatars for calls
- A significant percentage of test subjects mistook the AI for a human
- The percentage is around 48% or "nearly half"
Did outlets disagree about this?
Yes. Coverage of “Tavus unveils Griffin AI model, with nearly half of test subjects mistaking its video…” differs on 2 points. Each account below is how a different outlet described the same event:
Tavus's AI passed a Video Turing Test.
48% of test subjects mistook Tavus's AI for a human.
Which outlets covered this?
All 7 articles found on this story, grouped by the stance of the piece. Every link goes to the original publisher.