Runway’s AI video generator was trained on thousands of scraped YouTube videos

According to Runway, the company trained its video text generator using artificial intelligence on thousands of YouTube videos and pirated movies. report from 404 Media.And training data spreadsheet The materials obtained by the portal include links to YouTube channels belonging to major entertainment companies such as Netflix, Disney, Nintendo and Rockstar Games, as well as creators such as MKBHD, Linus Tech Tips and Sam Kolder.

There are also links to channels belonging to news agencies such as Edge, Novel Yorker, Reuters AgencyAND Wire“The channels in that spreadsheet were a company-wide effort to find high-quality videos to build the model from,” says a former Runway employee 404 Media“This was then used as input for a massive web crawler that downloaded all the videos from all of these channels, using proxy servers to avoid being blocked by Google.”

Runway is an artificial intelligence startup that received millions in funding from Google parent company Alphabet and Nvidia. It has created impressive tools that allow users to create realistic-looking AI videos, as well as ones that capture a specific type of animation. The latest tool, Runway, Gen-3 Alphalaunched in June and can “create videos in any style you can imagine.” Like other AI models, the Gen-3 Alpha must consume a wide range of content during training.

In addition to YouTube channels, 404 Media also found that the Runway dataset contains links to piracy sites like KissCartoon, which let you watch anime and other animated content for free. It’s still unclear whether Runway used all the movies in that spreadsheet to train its Gen-3 Alpha model — and we may never know. interview with TechCrunch In June, Runway co-founder Anastasis Germanidis said the company uses “curated, internal datasets” to train its models, but did not provide further details.

In response to a request for comment, Google indicated: Edge to the statement by YouTube CEO Neal Mohan, who said Bloomberg in April, it said training AI through videos on the platform was a “blatant violation” of its policies. Edge reached out to Runway for comment but did not immediately receive a response.