Runway Faces Criticism Over Allegations of Using YouTube AI Video Training Data Without Permission

Runway, a startup funded with hundreds of millions of dollars and backed by Google, is facing backlash from content creators following a report by 404 Media. The report cites a leaked spreadsheet allegedly revealing that Runway aimed to collect data from thousands of YouTube videos.

According to 404 Media, a former employee shared a spreadsheet showcasing Runway’s plans to categorize, tag, and train on YouTube channels belonging to major media companies like The New Yorker, VICE News, Pixar, Disney, Netflix, and Sony. This data reportedly contributed to a product called “Jupiter,” identified as Runway’s Gen-3 AI video creation model.

The spreadsheet also includes individual creators with large followings, such as Casey Neistat, Sam Kolder, Benjamin Hardman, and Marques Brownlee.

We have contacted Runway to verify the authenticity of the leaked spreadsheet and will provide updates as they become available.

Controversy Surrounding Gen-3 Alpha

Last month, Runway introduced Gen-3 Alpha, an earlier iteration of its software, which received praise for its realism. They opened it to public users a few weeks ago.

404 Media released a redacted version of the leaked spreadsheet, revealing over 3,900 YouTube channels and columns featuring various hashtags related to their content. One tab labeled “highcameramovement” lists more than 177 distinct YouTube accounts.

Creators and Critics Respond

While 404 Media indicated it couldn't confirm whether every video in the spreadsheet was used to train Gen-3, the existence of this data collection has raised concerns among creators and critics of generative AI. The implications of Runway potentially analyzing or replicating content without creator consent have sparked significant disapproval.

Tech reviewer Marques Brownlee (MKBHD) reacted on X with a skeptical comment, reflecting his previous criticisms of AI companies utilizing his videos without permission, despite his enthusiasm for AI technologies like OpenAI’s Sora.

Ed Newton-Rex, founder and CEO of ethical AI certification startup Fairly Trained, highlighted notable names from the spreadsheet, including channels belonging to musician Taylor Swift and filmmaker Wes Anderson.

YouTuber Omni, known as “Lay It Omni,” declared the spreadsheet “INSANE” and accused Runway of theft. Even AI filmmakers who have previously used Runway's tools, like Dustin Hollywood, are now voicing concerns about what they view as intellectual property theft.

While these criticisms are valid, it’s important to note that many companies, including notable names like Apple, Nvidia, and Anthropic, have faced similar accusations of using copyrighted materials without permission in training their models. A recent Wired article co-authored with Proof News highlighted these issues.

Legal perspectives on data scraping and training are complex; precedents exist, such as Google's practice of scraping the web for indexing. However, Runway is among multiple AI firms being sued by creators for unauthorized use of their work. In terms of public perception, Runway appears to be experiencing significant pushback following these revelations.