Amazon has been hit with a lawsuit by
content creators who allege that the company wrongly scraped their YouTube videos in order to train the generative artificial intelligence system Nova Reel, which aims to create videos from text
prompts.
"Rather than seek permission or pay a fair price for the audiovisual content hosted on YouTube, defendant harvested content creators’ protected and copyrighted
videos for commercial use and at scale without consent or compensation to the content creators," Ted Entertainment (which operates YouTube channels including h3h3 Productions), Matt Fisher (who
creates golf videos that he posts to the YouTube channel MrShortGame Golf) and Golfholics allege in a class-action complaint filed late last week in Seattle federal court.
They claim that
Amazon violated the Digital Millennium Copyright Act by circumventing digital locks on the YouTube videos in order to scrape them. That law prohibits people from getting around technology meant to
prevent them from copying digital media.
advertisement
advertisement
"On information and belief, Defendant used automated video downloading programs combined with virtual machines that rotated IP
addresses to avoid detection and blocking, enabling the mass unauthorized access and extraction of videos at the scale necessary to train Nova Reel," the complaint alleges.
The
complaint notes that Amazon says it trains Nova Reel on "curated data from a variety of sources, including licensed
and proprietary data, open source datasets, and publicly available data where appropriate."
But the plaintiffs argue that even if YouTube clips are publicly available, Amazon
doesn't have the right to harness them to train its model.
"The fact that a video can be accessed and viewed by the public through a web browser does not mean it can be
accessed, downloaded, extracted, and fed into a commercial AI training pipeline," the complaint alleges.
"Defendant cannot hide behind the word 'public' to launder what was in
reality an unauthorized mass unauthorized access and extraction of protected content."
An Amazon spokesperson declined to comment, other than to say the company doesn't discuss
matters in active litigation.