Tavus unveils Griffin, the “first Human Interaction Model”, which it says passed the “video Turing test”, with 48% of users thinking it was human in live chats
Introducing Griffin, the first model to pass the video Turing test. 48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video. It's the first Human Interaction Model (HIM).
@tavus
Context & Ripple Effects
Griffin sits within a move from generated avatars toward systems designed for continuous, real-time exchange: Thinking Machines Lab’s interaction-model preview framed responsiveness and collaboration as a distinct model capability. Earlier, Google Research’s VLOGGER work focused on producing lifelike speaking and gesturing video from a photo and audio clip.
Tavus’s reported live-chat result shifts the competitive claim from visual realism alone to whether an interlocutor can distinguish an AI during an exchange. Public reaction focused heavily on impersonation and companion-like harms, concerns expressed by commenters rather than established outcomes.
First-order effects
Tavus gains a concrete, if company-reported, differentiation claim for Griffin: 48% of live-chat participants believed they were speaking with a human, alongside its top ranking on NVIDIA’s full-duplex AI-video benchmark.
People interacting with Griffin in the tested format face a material risk of misidentifying the system, making disclosure and identity cues immediately more consequential for Tavus deployments.
Second-order effects
Providers of full-duplex video agents must compete not only on latency and avatar quality but also on credible evidence of how their systems behave in live human conversations.
Enterprise buyers considering video-facing AI will have stronger reason to require explicit bot labeling and escalation paths, particularly as public discussion frames human-like interaction as an impersonation risk.
Third-order effects
If live video agents become difficult to distinguish from people at scale, anthropomorphic-AI governance is likely to center on identity disclosure and safeguards against deceptive interaction rather than on video generation quality alone.
The trend: AI interfaces are moving from producing humanlike media to sustaining humanlike real-time interaction, increasing the importance of disclosure and trust controls.
48% is bullshit, but I cannot think of any reason this needs to exist outside of scamming old people or driving lonely people to psychosis with fake relationships
thrilled to announce we have built the “trick old people into handing over their bank passwords machine” from the classic sci-fi horror novel “don't build the trick old people into handing over their bank passwords machine”
i hope the people building stuff like this are someday kept up at night by what they're taking away from the entire world in exchange for money. 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.
In a new experiment, nearly half the people on a video call thought they were speaking to a human. They weren't. It was a new AI model. That is INSANE. Does anyone in America think it is a good idea to race forward to build machines that replace us? PAUSE AI NOW.
This is a disaster for the human race and I've really lost patience with the useless morons who can't see that. An AI video chat that passes for human has no necessary legitimate applications. It will be used to scam old people out of their life savings. Lonely, mentally vulnerab…
Yesterday we previewed Griffin, a Human Interaction Model capable of seeing, hearing, sounding, and looking like a human does. There have been a lot of questions, so I wanted to take a moment to share our thoughts. By way of introduction: Tavus is a research lab focused on enabli…
Why are companies selling fake human beings Why does the computer software we interact with have to be indistinguishable from a living, breathing person Why is this an achievement
Stop. Building. Tech. To. Trick. Humans. 👏 Those who chase money over what is right for humanity or healthy for society should not be rewarded. Happy to engage with the counterposition. What actual good comes from this?
We should not be celebrating technology that can deceive people and blur the distinction between what's real and what's fake! It should be illegal — plain and simple — for any AI to impersonate a human being without a clear and visible disclaimer that it is, in fact, AI.
The whole point of building Tavus has been simple: talking to a machine should feel as natural as talking to a friend or coworker. It's hard to describe all the tiny nuances that make a conversation feel human. The little expressions. Moving around in your chair. Knowing when to …
This is actually insanely dangerous for the loneliness epidemic... AI girlfriends are going to absolutely cook society. There are millions of dudes who already get no matches, no dates, no attention and have basically given up. Now give them a hot girl on FaceTime who looks real,…
sigh.. must say a few things about this is face to face video chat the most efficient way to communicate with a machine? imagine you want to cancel your netflix subscription. would you prefer a button click, or turn on your camera and talk to a bot face to face? how about booking…
Because it's the first model that can be mistaken for a real person, Griffin requires safety considerations before it can be released publicly. The full research preview covers how it works, how we tested it and the safety measures that have to come first: https://www.tavus.io/gr…
While Mars is still telling Griffin about a boring date, it laughs, nods and asks questions instead of waiting for her to stop. That's full duplex: listening, watching and talking at the same time, the way people actually talk. [video]
When machines meet us where we are and truly understand us, we unlock a future where using a machine will be as easy as talking to a friend or coworker. A tutor for every student that notices when they're lost, an elderly care companion that listens, or the perfect assistant. [vi…