Tavus Previews Griffin, a Live AI Video Model That Testers Mistake for Human
Summary
Tavus previews Griffin, a live AI video model that sees, hears and reacts in real time, with 48% of Griffin-Lite testers believing they are speaking to a human; access remains limited to trusted testers pending safety and disclosure features.
Key Points
- Tavus previews Griffin, a live-video Human Interaction Model that sees, hears, speaks, and reacts, with 48% of Griffin-Lite test participants believing they are talking to a human.
- Griffin incorporates visual details from users’ screens and responds mid-conversation with behaviors such as nodding instead of waiting for a turn.
- Tavus limits Griffin-Lite to trusted testers while it develops safety and disclosure features, with a public rollout planned later.