Entertainment · Books & Arts · 5 hrs ago
Artists turn videos from retired Google AI model into new work
“Post Generation” is an exhibition at tiat gallery in San Francisco.
Three artists—Zelia ZZ Tan, Karyn Nakamura and Haley Mortin—made new work in response to videos produced by Veo 1, a Google text-to-video model.
Veo 1 was used internally in 2024 and later leaked, but it is no longer available.
It made only six-second videos, often with surreal images, and researchers generated hundreds of thousands before it was shut down.
Google replaced it with Veo 2 and Veo 3, which were released to the public.
The old model’s outputs are not publicly available, so organizer Tina Tarighian shared part of its archive with the artists.
Their exhibition uses those videos to reflect on AI technology and preserve a record of an earlier stage in its development.
An exhibition at tiat gallery in San Francisco gives artists access to videos generated by Google’s retired Veo 1 model.
The exhibition, titled “Post Generation,” features new work by choreographer Zelia ZZ Tan, creative engineer Karyn Nakamura and fiber artist Haley Mortin.
Veo 1 was developed in 2024 and could generate only six-second videos, often with surreal results.
Researchers generated hundreds of thousands of videos before the model was shut down and replaced by Veo 2 and Veo 3.
Google creative technologist Tina Tarighian shared part of the archive with the artists to explore what could be made from the model’s remaining outputs.
- Who
- Artists Zelia ZZ Tan, Karyn Nakamura and Haley Mortin created work using Veo 1 videos. Tina Tarighian organized the exhibition.
- What
- Post Generation” presents new work responding to an archive of videos generated by a retired Google AI model.
- When
- The exhibition is described in an article published October 12, 2026. The exhibition dates are not stated.
- Where
- tiat gallery in San Francisco.
- Why
- Tarighian wanted to explore what artists could make from the archive of a model that could no longer generate videos and whose outputs were not publicly available.
This story does not have two clearly opposing sides.
Early text-to-video models produced videos of Will Smith eating pasta with visibly distorted results.
Google developed Veo 1, a model that was released only for internal use and later leaked.
Researchers generated hundreds of thousands of videos with the model.
Veo 2 and Veo 3 succeeded it, and Tarighian shared part of its video archive with three artists for “Post Generation.
- Exhibition
- Post Generation
- Venue
- tiat gallery, San Francisco
- Retired model
- Veo 1, developed by Google in 2024
- Video length
- Six seconds per video
- Artists
- Zelia ZZ Tan, Karyn Nakamura and Haley Mortin
- Archive size
- Researchers generated hundreds of thousands of videos
What people are saying
No one can generate using Veo1, and the outputs it produced aren’t publicly available
So I thought, what can we do with these leftover scraps of a model? Can we gift this archive to artists?
I thought it was just a data entry job
Sources
Related news
Sacha Baron Cohen to star in Tarantino’s first West End play
A Jammed Conveyor Belt Tests an Accessible Art Installation
Flor Khan’s Bronx Studio Connects Art With Family Memories
Christie’s to Auction Barbara Hepworth Sculpture in London on October 21
Made of Chennai Kolu Kondattam 2026 contest opens for entries
Morante returns in a bullfight that drew limited excitement
Nina turns a Chekhov audition into a study of power and consent
Anne Carson wins 2026 Nobel Prize in Literature
Tarek Atoui fills Tate Modern’s Turbine Hall with sound and sculpture
Morante faces hostility on return to Madrid’s Las Ventas bullring
Rosana Paulino honored in New York for work challenging scientific racism
