The uncanny valley just got a footbridge. tavus released griffin, a conversational video model, on…
the uncanny valley just got a footbridge. tavus released griffin, a conversational video model, on oct 8, and its own study found 48% of callers believed they were talking to a human.
the turing test is now being scored by people on the phone, not judges in a room.
Context
Tavus's Griffin page says that 48% of people believed Griffin was a real person after a one-minute video call, and that its previous systems had a maximum 2% pass rate. Decrypt reports that 26 of 54 participants believed Griffin-Lite was a real person, versus 1 of 41 for the previous system, and that the results are the company's own.
Decrypt says Griffin-Lite ranks first on Nvidia's VideoFDB benchmark with 3.83 out of 5 on generation against 2.80 for the next-best system and 3.92 for a human reference. It says participants were told they would meet another person, and that Griffin-Lite is limited to select trusted testers while Tavus works on safety measures.
The 48% matches Tavus, and it is 26 of 54 participants, a small sample from the company's own study. The note says callers, and the study says participants who were told they would meet another person, so it is a lab test and not an audit of phone callers.
The note says Tavus 'released' Griffin on October 8. Decrypt and RuntimeWire date the unveiling to October 1, 2026, and describe a research preview limited to select testers, so a release on October 8 is unsupported here, not refuted.
'The model that passed the Turing test' framing is Tavus's own: it calls Griffin the first model to pass the real-time video Turing test. 'Scored by people on the phone, not judges in a room' is the author's line, and the study used a one-minute video call.
Related work
- Griffin: The First Human Interaction Model (Tavus) ↗Tavus's page with the 48% claim and the previous pass rate.
- This AI Is Already Fooling People on Video Calls Into Thinking Its Human, Company Says (Decrypt, October 2, 2026) ↗Independent report with the sample size, the benchmark and the limited access.
- Tavus says Griffin fooled 48% of testers in one-minute video calls (RuntimeWire, October 1, 2026) ↗Report dated October 1 with the sample counts.
Watch next
- Read Tavus's research page for the study protocol. Look for independent replication or a wider release.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 9 October 2026 at 14:31 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →