Video & Creative AI
Tavus Griffin passes human test in live video calls
The Rundown AIhigh confidence
Why it matters
- →Real-time visual grounding in video calls marks a significant leap over static avatar technology.
- →High deception rates (48%) highlight urgent ethical risks for identity verification and fraud prevention.
- →The shift from passive presentation to active conversation changes the utility of digital humans.
AI video startup Tavus previewed Griffin, a model capable of real-time visual and auditory interaction during live video calls. In controlled studies, nearly half of users failed to distinguish the AI from a human counterpart, with performance metrics closely matching real people on NVIDIA's VideoFDB benchmark. The system processes visual cues and screen content dynamically rather than relying on scripted responses. Tavus is currently restricting access to trusted testers while developing safety disclosures.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1OpenAI Unveils GPT-Live for Voice Interaction
OpenAI · July 8, 2026 · Background
- 2Google Vids Enhances Video Creation with AI Tools
Google AI Blog · July 16, 2026 · Background
- 3Google Vids Adds AI Avatars and Gemini Omni
TechCrunch AI · July 16, 2026 · Background
- 4OpenAI's GPT-Live Enhances Voice AI Interaction
OpenAI · August 3, 2026 · Background
- 5Google DeepMind Launches Agentic Video Understanding
Google DeepMind · September 1, 2026 · Background
- 6AI 'Actor' Tilly Norwood fails basic reality checks
WIRED AI · September 15, 2026 · Background
- 7Google launches Gemini 3.8 Live Avatar for enterprise
The Verge AI · September 24, 2026 · Related
- 8Synthesia launches interactive AI avatars for enterprise roleplay
TechCrunch AI · September 26, 2026 · Background
More in Video & Creative AI
© Matt WolfeVideo & Creative AImusic
ElevenLabs Releases Eleven v4 and Microsoft Launches MAI Voice
ElevenLabs launched its v4 voice model for higher fidelity audio, while Microsoft introduced MAI-Voice-2.1 and a new streaming transcription model.
Matt Wolfe
© MIT News AIVideo & Creative AIother
InstructMesh repairs AI 3D models for real-world use
Most generative 3D models look good but fail structurally, leaving novices unable to fix them. InstructMesh solves this by combining Microsoft’s TRELLIS generator with GPT-4 reasoning, allowing users to describe flaws in natural language and get corrected geometry instantly. MIT researchers proved that non-experts could identify and repair nearly 90% of structural errors in existing models using the tool. This shifts 3D fabrication from a technical hurdle to an intuitive design process where visual intent matches physical function.
MIT News AI
© TechCrunch AIVideo & Creative AIimage
ChatGPT launches global virtual try-on and favorites
OpenAI is pushing ChatGPT deeper into e-commerce with a global virtual try-on feature powered by its new Images 2.5 model. Users can now upload selfies or product screenshots to visualize clothing on themselves, moving beyond simple text-based recommendations. This update also introduces a Favorites library for saving discoveries, directly challenging Pinterest and Google’s dominance in visual fashion search. While the underlying image generation improvements are notable, the real shift is ChatGPT becoming a proactive shopping assistant rather than just an information retriever.
TechCrunch AI