Tavus' real-time video model Griffin fooled 48% of participants in human test
thione · x · 2026-10-06
Tavus introduced Griffin, a real-time video-to-video "Human Interaction Model" that transforms live video into a photoreal driven persona.
- In Tavus' own small company study, 48% of participants believed they were talking to a human, showing real-time visual realism is nearing the uncanny-valley threshold.
- The model targets video conferencing and avatar interaction, but also raises deepfake and identity-impersonation concerns.
More from Multimodal
- AI-Generated 'Tour of Armenia' Video With Muscular Guide Amuses Reddit — EternalSnow05 · 2026-11-19
- One image, 5x price gap: Gemini 4 Neon costs $1.01 vs GPT-6.2 Sol at $0.19 — Angaisb_ · 2026-11-19
- Creator Shares Opening Scene of '23:17', an 8-Episode AI Mystery Anime Series — SimMonzi · 2026-11-19
- Gemini 4 Neon (xhigh) demo: generating a clean Ana de Armas portrait as SVG — lyraxana · 2026-11-19
- Infinidream 0.17.2 brings 1080p Electric Sheep and screen-as-installation modes — scott_draves · 2026-11-19
- Fitting 5 character LoRAs in one image: 4 methods tested, best hits 3/5 — PoleRacers · 2026-11-19