Tavus' Griffin AI avatar fools 48% of users in one-minute call
Tavus has launched Griffin, its first Human Interaction Model, which processes speech, facial expressions, tone, gestures and pauses while generating and receiving video. In a Tavus study, 48 percent of participants believed Griffin was a real person after a one‑minute video call, compared with only 2 percent for previous systems.
Key points
- 48% of participants believed Tavus' Griffin AI avatar was a real person after a one‑minute video call.
- Griffin scored 3.83 on Nvidia's human‑likeness test, close to humans' 3.92, surpassing previous AI models' 2.80.
- Tavus, founded 2020, has raised about $64 million and is expanding from AI video sales tools to live conversational avatars.
An independent Nvidia test measured how human an AI feels in direct audio‑video conversation. Griffin scored 3.83 points, close to humans’ 3.92, and far ahead of the previous best AI model at 2.80. A preview called Griffin‑Lite is available to select testers as a research preview, with a more capable version coming once safety concerns are addressed.
Founded in 2020, Tavus has raised about $64 million. It began building personalized AI videos for sales and marketing before expanding into live video conversations with digital personas, citing tutoring, difficult‑conversation practice and camera‑based tech support as potential use cases.
Nearly half of test subjects mistook Tavus' AI video avatar for a real person on a one-minute call
The Decoder · 1 October 2026
Loading the full article…
This text was published by The Decoder and written by Matthias Bastian. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Google releases Gemini 4 Argon with 1M token output limit · 19 src
- Graphite finds 13,000 phrases distinguishing AI writing from human · 1 src
- Ideogram introduces Ideogram 4.5 for precise image editing · 1 src
- Google rolls out Gemini 4 Argon to trusted cyber defenders first · 15 src
- OpenAI releases GPT-6.1 Sol, says it nearly matches Astra on agentic coding at one-fifth the price · 35 src
Comments
via GitHub Discussions