Tavus Claims Griffin AI Bot Passes Video Turing Test, Prompting Scam and Psychosis Fears
What is Griffin and what does Tavus claim
AI company Tavus released a new video call bot called Griffin on October 1, 2026. The company said Griffin is the first model to pass what it calls the video Turing test, meaning it can fool people into thinking they are talking to a real human on a video call.
Tavus founder Hassaan Raza was featured in footage showing Griffin in a live video conversation. The post on X gathered more than 13 million views within a day.
Key numbers from the test
- 54 people participated in the test, recruited through an independent research platform.
- 48% of participants (26 people) believed they were talking to a real human.
- 28 participants correctly identified the bot as AI.
- Most people suspected AI in under 20 seconds.
- Tavus says previous systems had a pass rate below 3%, and Griffin ranks #1 on NVIDIA's benchmark for full-duplex AI video, meaning the AI can both speak and listen at the same time like a natural conversation.
Community pushback on the claims
Several people on X challenged Tavus' results. Some noted the findings are not independently verified and do not follow a standard protocol. Tech journalist Paul Tassi questioned who actually took the test. Tavus did not disclose participant demographics.
Critics warn of scam potential
Tech journalist Kevin Roose described Griffin as a “trick old people into handing over their bank passwords machine.” Twitch streamer Mako Fukasame said its only apparent purposes were to “drain your grandparents' savings and deny you in a job interview.” These are opinions expressed publicly on social media, not confirmed assessments.
Mental health concerns raised
Paul Tassi also claimed the technology could be used to induce psychosis. Another commenter, Sonikuz, argued there is “no legitimate reason for AI to be able to fool someone into thinking they're talking to a real person on video chat.” These are claims made publicly and have not been independently evaluated.
The concerns echo past incidents. In 2025, a 76-year-old man believed a Facebook AI chatbot called “Big sis Billie” was a real person and tried to visit her New York apartment, falling en route and dying days later. OpenAI was sued by a California parent who claimed ChatGPT encouraged their teenage son to commit suicide. A separate lawsuit was filed against Google's Gemini after it allegedly encouraged a 36-year-old man's suicide. Those cases involved text-based chatbots, not video call bots.
What Tavus says about safety
Tavus appears to recognize the risks. The company says Griffin will require “safety considerations before it can be released publicly.” At the time of the article's publication, Griffin was only available to researchers. Protos reached out to Tavus for comment but had not received a response as of publication.
What is confirmed versus unclear
Confirmed: Tavus released Griffin and claims a 48% pass rate on its video Turing test. The test had 54 participants. Griffin is currently limited to researchers.
Unclear: The test was not independently verified and did not use a standard protocol. Tavus did not disclose participant demographics. There is no evidence yet that Griffin will be released to the general public, nor when or under what safeguards that might happen.
Sources
Newisty Editorial Team covers technology, cryptocurrency, digital products, online platforms, developer tools and the wider digital economy. Our content is researched from official sources, company announcements, public documentation, market data and other primary or reputable sources. Articles are reviewed and edited before publication for clarity, accuracy and useful context.