Online streamers sue Twitch, Amazon over generative AI training

The proposed class of video streamers claims their content is being used to train Amazon’s AI models without permission or compensation. A sign with Twitch Interactive's logo sits above the entrance to the company's headquarters in San Francisco, CA on February 15, 2025. (Matt Simons/Courthouse News) SAN FRANCISCO (CN) — A Connecticut-based Twitch streamer sued the popular video streaming platform and its parent company Amazon in a class action over claims the tech companies harvest creators’ video streams without permission or compensation to train Amazon’s generative artificial intelligence programs. Plaintiff Warren Pandiscia, a Twitch content creator with more than 900 followers, claims Twitch and Amazon unlawfully copied millions of videos to develop Amazon’s AI models and products, violating creators’ implied contracts with Twitch and state unfair competition law. “Because Amazon AI products are commercialized, Amazon had an overwhelming incentive to acquire training data on an unprecedented scale. Rather than negotiate for lawful licenses or seek permission, defendants accessed the Twitch streams and videos to utilize them as a massive dataset necessary to fuel Amazon’s AI products,” Pandiscia said in his 37-page complaint filed in the Northern District of California Thursday. Twitch confirmed its platform uses creators’ content for AI training through a post on X, formerly Twitter, on Aug. 12. “We’ve added a setting that lets you opt out of having your channel content used to train generative AI content models across Amazon,” the post read, linking to an FAQ page. That same day, Twitch updated its Terms of Service and Privacy Policy with new language on automated tools and artificial intelligence. According to Twitch’s FAQs , Amazon may use streams, past broadcasts, clips, chats, pictures and text for AI training, including “future training of a model developed by Amazon whose purpose is to generate or synthesize text, audio, images or video.” The page provides an example of how their data may be used: users’ audio could help “refine models that create speech to text, which would help improve captions at Twitch but would also help improve captions across Amazon.” Twitch channels are automatically opted in to AI training unless users manually opt out. However, settings apply by channel, so content from an opted-out user who engages on an opted-in channel may still be used. “By design, defendants never obtain—and their systems are incapable of obtaining—the consent of all parties to the communications they capture,” Pandiscia says. Pandiscia said he had “no reason to believe” Twitch and Amazon would use his content to train generative AI products and is “opposed to such use of his original content.” He said he opted out immediately after learning of the setting and would have done so earlier had it been disclosed. Pandiscia says the companies’ actions cost him money and property and deprived him of control over his work. He claims they scraped streamers’ content without permission as early as 2024 and that Twitch broke commitments to obtain consent before making policies “less protective of personal information” and to notify users before sharing their information with third parties. “Content creators such as plaintiff and the class members will never be able to claw back the intellectual property unlawfully copied and used by defendants to train Amazon’s generative AI,” he said. Pandiscia asserts claims of breach of implied contract, unjust enrichment, breach of express contract and unfair business practices. He asks the court to award injunctive relief, as well as damages, restitution and disgorgement of profits. Representatives for either party did not immediately respond to a request for comment. In 2024, it was reported that Mike Minton, then Twitch’s Chief Monetization Officer, acknowledged that Twitch content was already being used for AI development “in a prototyping, not in any kind of production scale, capacity