Twitch and Amazon are facing a proposed nationwide class action over allegations that creator content was used to train generative artificial intelligence models without adequate consent or compensation.
Connecticut streamer Warren Pandiscia filed the lawsuit on August 20, 2026, in the U.S. District Court for the Northern District of California. The case is listed as Pandiscia v. Twitch Interactive, Inc. et al., No. 3:26-cv-08721. Pandiscia has requested a jury trial.
The 37-page complaint alleges that Twitch and Amazon used streams, videos, clips, chat messages and other channel material to develop and improve generative AI products. The complaint brings claims including breach of contract, breach of the implied covenant of good faith and fair dealing, unjust enrichment and violation of California's Unfair Competition Law.
These allegations have not been proven in court. The proposed class has not been certified, and the defendants have not yet filed a response to the complaint.
Twitch's AI Training Setting
The dispute follows Twitch's August 12 announcement that creators could opt out of having their channel content used to train generative AI models across Amazon.
The setting is available under Twitch's account settings in the Security and Privacy section. Twitch's documentation says eligible material can include streams, VODs, clips, highlights, stream chats and text or images posted to a channel.
The setting is enabled by default, meaning creators must manually turn it off if they do not want eligible content from their own channels to be included in future training covered by the feature.
The announcement prompted criticism from streamers who said they had not been clearly asked to provide permission. It also raised questions about whether Amazon had used Twitch material before the opt-out setting was introduced.
During a livestream about the policy, Twitch Chief Product Officer Mike Minton was asked why the feature was opt-out rather than opt-in. He responded, "If this was opt-in, nobody would opt in," according to reporting and a clip shared by streaming reporter Zach Bussey.
Minton also did not provide a definitive answer about which Twitch content Amazon may have used for training in the past.
What the Lawsuit Alleges
Pandiscia's complaint argues that Twitch's earlier agreements did not clearly authorize the use of creator content to build separate commercial generative AI products.
The filing also focuses on changes to Twitch's terms and privacy disclosures made around August 12. According to the complaint, the updated language gave Twitch and its affiliates broader rights involving creator content.
Pandiscia argues that the timing is significant because the policy change and public disclosure of AI training occurred at the same time. Twitch and Amazon may argue that the updated language clarified existing rights rather than creating new ones.
The complaint alleges that Twitch content may have been used for AI development as early as 2024. That claim remains part of the plaintiff's allegations and has not been established by a court.
Opting Out Does Not Remove Earlier Training Data
Twitch's opt-out setting applies to future generative AI training covered by the feature. It does not automatically remove content that may already have been collected or used in a previously trained model.
The setting also applies differently to chat messages posted in other channels. Twitch says that the preference of the channel where the chat appears determines whether that chat content is eligible for training. A creator's own channel setting does not necessarily control messages posted elsewhere.
The opt-out also does not stop every use of content or data on Twitch. Other platform functions, including certain safety, recommendation or support tools, may be governed by separate policies.
The Role of Amazon Nova Reel
The complaint references Amazon Nova Reel and Amazon's broader work involving large video datasets. However, the filing does not establish that Pandiscia's streams, or Twitch content generally, were specifically used to train Nova Reel.
Instead, Nova Reel is cited as an example of the type of commercial generative AI product that may require extensive video data.
That distinction is important because the lawsuit remains at the complaint stage. A reference to an AI product in the filing is not proof that a particular creator's content was used to train it.
What Happens Next?
The case will proceed through the federal court process. Amazon and Twitch may challenge the complaint, seek dismissal or argue that individual arbitration or existing platform agreements limit the claims.
The proposed class also must be certified before it can proceed as a nationwide class action. Until that happens, the case represents Pandiscia's claims and does not establish that all Twitch creators were affected in the same way.
The lawsuit raises a broader question for online platforms: whether a creator's permission to upload, host and monetize content also allows a platform or its parent company to use that material to develop unrelated generative AI products.
For Twitch creators who do not want their eligible channel content used in future AI training, Twitch currently provides an opt-out setting under Settings → Security and Privacy → Training for Generative AI.
Meta description:
Twitch and Amazon face a proposed class action alleging creator content was used to train generative AI without proper consent or compensation.
Keywords:
Twitch AI lawsuit, Amazon AI training, Warren Pandiscia, Twitch creators, generative AI, Twitch opt out, creator rights, class action lawsuit
