EP 647 : Tavus Griffin AI Video Calls Fool 48% – Google Protein Watermarks, OpenAI & Microsoft

Imagine opening a video call and not being sure whether the person on the screen is human or an AI. In today’s episode we dive into Tavus’ groundbreaking Griffin model, which fooled 48 % of testers into believing they were speaking with a real person. We break down how Griffin‑Lite watches, listens, and interjects mid‑sentence, its near‑human score on NVIDIA’s VideoFDB benchmark, and the safety safeguards Tavus is building before a wider rollout.We also explore Google DeepMind’s new SynthID Bio, a proof‑of‑concept that embeds invisible watermarks into AI‑generated protein structures, promising traceability for custom DNA designs while keeping AlphaFold 3’s accuracy intact.Our AI news roundup covers Ideogram 4.5’s precision‑editing image model, OpenAI’s internal security shake‑up, a historic cipher cracked with GPT‑6 Astra, Microsoft’s MAI‑Voice‑2.1 and the record‑breaking MAI‑Transcribe‑2‑Streaming, and California’s “No Robo Bosses” law.Finally, we highlight today’s top productivity tools, from FLUX 3 Image’s box‑by‑box control to Shopify’s Canvas workspace, Omnia Agent for GEO tasks, Clarity’s speech‑enhancement, Twin the local‑first Mac assistant, and Vorker the AI‑powered small‑business coworker.Listen now, share with your tech‑savvy friends, and hit subscribe for weekly AI deep dives.Tools mentioned: Griffin‑Lite (Tavus), SynthID Bio (Google DeepMind), Ideogram 4.5, FLUX 3 Image (Black Forest Labs), MAI‑Transcribe‑2‑Streaming, MAI‑Voice‑2.1 (Microsoft), Canvas (Shopify), Omnia Agent, Clarity, Twin, Vorker.