Twitch and its parent company Amazon are facing a proposed class-action lawsuit over claims that content from Twitch streamers was used to train Amazon’s generative artificial intelligence systems without clear permission or payment.
The lawsuit was filed on August 20, 2026, in the U.S. District Court for the Northern District of California. It was brought by Twitch creator Warren Pandiscia on behalf of himself and other creators whose content was allegedly used for AI training. The case is still at an early stage, and the claims have not been proven in court.
According to the complaint, Twitch and Amazon allegedly used creator content such as livestreams, recorded videos, clips and other channel material to help train, improve and commercialize Amazon’s generative AI products. The filing argues that creators did not knowingly agree to this type of use and were not compensated for providing training material.
The dispute became more important after Twitch introduced a new “Training for Generative AI” setting on August 12. The option allows creators to stop their channel content from being used for future generative AI training across Amazon. However, the setting was enabled by default, meaning creators who do not want their material included need to manually opt out.
That default approach has raised concerns among streamers because Twitch content can contain much more than gameplay. Livestreams often include a creator’s voice, face, commentary, conversations, original graphics and hours of unique video. Recorded broadcasts, clips, chat content, images and text may also provide valuable data for companies building AI systems that understand or generate different types of media.
The lawsuit goes further by alleging that Twitch creator material may have been used for Amazon AI development before the new opt-out option appeared. The complaint claims that such activity may have taken place as far back as 2024. Whether those allegations can be proven will likely become one of the key questions in the case.
The complaint also points to Amazon’s growing investment in generative AI, including systems capable of creating video from text and images. The lawsuit argues that large collections of creator-made video can have significant value when companies need huge amounts of Training Data to build and improve these types of models.
Legally, the proposed case focuses on issues including breach of contract, unjust enrichment and California’s unfair competition law. The plaintiff argues that the permissions Twitch creators gave the platform did not clearly authorize their work to be used to train separate commercial AI products.
For Twitch creators, the lawsuit highlights a wider problem emerging across the technology industry: what exactly users agree to when they upload content to large online platforms. As AI companies search for more text, audio, images and video to improve their models, creators are increasingly questioning whether older platform agreements provide enough permission for AI training.
The case could therefore become important beyond Twitch. A ruling on how platform agreements apply to AI training could influence how social networks, video platforms and other online services explain the use of creator content in the future.
For now, the lawsuit remains a proposed class action, and no court has ruled that Twitch or Amazon acted unlawfully. But the dispute adds to the growing pressure on technology companies to give creators clearer choices about whether their work can become training data for commercial AI systems.
