A Twitch creator has filed a proposed class-action lawsuit against the Amazon-owned streaming platform for allegedly using the content of millions of its creators to train Amazon’s AI products without their consent. The lawsuit was filed a little over two weeks after Twitch notified users they were required to opt out if they did not want their content used to train Amazon’s generative AI content models. The announcement confirmed that, without an opt-out, eligible channel content could be used to train future Amazon generative-AI models. But according to the complaint, Amazon had used creator content for AI development before the announcement.
Twitch is a live-streaming platform Amazon acquired in 2014 for $970 million. According to the lawsuit, Twitch had roughly 240 million active monthly users by 2026, with 26 to 30 million people visiting the platform each day to view content from 3.2 million to 6.9 million creators each month. The creators can earn money from their broadcasts through subscriptions, ads, and tips.
The class-action lawsuit against Twitch and Amazon was filed Aug. 20 in the U.S. District Court for the Northern District of California in San Francisco. The named plaintiff is Warren Pandiscia, who filed the suit on behalf of himself and a class that likely includes millions of individuals. Pandiscia is a Connecticut resident who has regularly created Twitch content for 10 years, and his channel has more than 900 followers.
The lawsuit claims Twitch and Amazon's use of creator content to train AI constitutes abuse and exploitation of the content creators’ work because it was not authorized. It also contends the two companies derived substantial benefit from infringing on creators' video and streaming content. Creators invest time and resources creating original content for Twitch that Twitch and Amazon exploit without their consent, the suit says.
“Because Amazon AI products are commercialized, Amazon had an overwhelming incentive to acquire training data on an unprecedented scale,” the complaint explains. “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.”
Option to Opt Out of AI Training Given in August
On Aug. 12, Twitch confirmed that Amazon would use its content for AI training and that users who did not want their content used for that purpose would need to opt out of the program. The lawsuit said the opt-out requirement shows that Twitch’s default configuration allows Amazon to collect live videos, recorded videos, chats, and other creator content for AI training. To opt out, streamers must go to their account options and disable the “Training for Generative AI” option.
The lawsuit says the defendants knew Twitch creators largely oppose their content being used to train AI. But instead of telling the Twitch community that Amazon is using the platform’s content for AI training, the defendants framed the August announcement as adding a setting that allows users to opt out of having their content used to train generative AI models. Twitch’s FAQ page on account settings specifies that users are only opting out of future training and AI model development.
According to the lawsuit, it appears Twitch and Amazon may have been mining creator content to train AI as far back as 2024. The terms of service in effect for Twitch since March 2024 stipulated that users grant Twitch and its subsidiaries the right to use, reproduce, adapt, distribute, and create derivative works from their content, it said. However, until the August announcement of the opt-out, Twitch had not explicitly said it would use user content to train generative AI models.
During a live question-and-answer session, Twitch Chief Product Officer Mike Minton responded to a question regarding whether creator content was already being used to train AI by saying that he did not know what Amazon has and has not used for AI training, the lawsuit claims. He also allegedly defended the use of an opt-out provision, saying nobody would opt in to having their content used for AI training.
Causes of Action
The lawsuit claims four causes of action:
- Breach of implied contract and the implied covenant of good faith and fair dealing. By using Twitch’s platform and participating in its revenue-sharing program, the creators claim they entered into an implied contract. Under that implied contract, Twitch would use creators’ content only within the scope of the disclosed relationship and would not copy it at scale for undisclosed commercial exploitation outside the relationship.
- Unjust enrichment/quasi contract/restitution. The suit argues that Twitch and Amazon received benefits from content creators and unjustly retained them. Specifically, the plaintiffs conferred a direct economic benefit on the defendants by providing original content. That transaction included an implicit commitment that the original content would not be used for commercial gain without authorization.
- Breach of express contract. Twitch and the creators entered into express contracts with the stated purpose of exercising rights in connection with monetizing Twitch’s services. The creators maintain that they complied with the agreement, but the defendants violated the contract by using creator content for reasons other than monetizing Twitch’s services and to build Amazon’s separate generative AI products.
- Violation of California’s Unfair Competition Law. The suit claims Twitch and Amazon’s practices were unlawful violations of the Unfair Competition Law by secretly capturing creators’ works and communications for undisclosed commercial AI training and using all of the creators’ work for AI training by default. It also maintains the defendants’ practices were likely to deceive reasonable members of the public.
What This Lawsuit Could Mean for Twitch Creators
The lawsuit could help answer an important question for Twitch creators: Do a platform’s general terms of service give it permission to use streams, videos, chats, and other creator content to develop generative-AI products?
The defendants will likely argue that creators consented to the use of their content, even if the terms of service did not explicitly say it would be for AI training purposes.
If the case moves forward and the creators prevail, Pandiscia is seeking relief that could include money damages, restitution or disgorgement of profits, and an order limiting or stopping the alleged use of creator content for AI training.
