Twitch and Amazon are being sued in a class action lawsuit regarding the training of AI using livestreams.

      On August 20, a Twitch streamer filed a lawsuit against Twitch and Amazon regarding the use of livestreams to train Amazon's generative AI. The 37-page complaint was submitted in the Northern District of California, San Francisco Division, under case number 3:26-cv-08721.

      The primary issue raised in the lawsuit involves a contract. On August 12, Twitch announced the introduction of an opt-out feature. On the same day, it updated the relevant clause in its Terms of Service related to the use of a creator's content.

      The clause altered on August 12

      Previously, Section 8(a)(i) granted Twitch a license for a specific purpose. The complaint cites the former language: "Twitch can exercise these rights in connection with monetizing the Twitch Services." The updated language now states “in connection with Twitch and/or its affiliates’ business," as per the complaint, with Amazon being considered an affiliate.

      The amendment also introduced a new provision, stating that by streaming, a user is “directing Twitch to also cache and/or store your User Content and every individual piece of it.” The complaint notes that no similar language existed in earlier versions and contends that these changes "are themselves evidence" that past terms "did not convey the rights Defendants exercised."

      What the previous terms indicated regarding deletion

      The complaint references section 8(a) concerning the creator’s options when removing content. It states that the license “terminate[s] once you delete such User Content from the Twitch Services or generally by closing your account.” According to the complaint, AI training is not listed among the activities that remain post-deletion.

      It refers to two additional Twitch policies: Section 7 prohibits “use of any data mining, robots, or similar data gathering or extraction methods,” and Section 10 restricts access “by means of any robot, spider, scraper, crawler, or other automated means.”

      Previous privacy notice details

      The privacy notice, last updated on January 15, 2026, listed how Twitch utilized user information, with no mention of AI training, the complaint states. On August 12, Twitch modified the wording, which now states that the company uses information “to test and improve technology, including machine learning and artificial intelligence,” and adds that data may be utilized “to develop or deploy our generative AI models and services.”

      Both versions contain two promises: that Twitch would not make its policies “less protective of personal information collected in the past” without consent, and it promised users notice before sharing their information with third parties.

      Who the plaintiff is

      Warren Pandiscia, a resident of Fairfield County, Connecticut, has been streaming on Twitch for nearly a decade, with over 900 followers and more than 1,000 hours of content. The filing states he has expended tens of thousands of dollars on equipment, props, and subscriptions, focusing on building LEGO sets and playing video games. He opted out upon learning about the new setting. He is represented by Almeida Law Group and Sterlington.

      Comments from Mike Minton

      The complaint references statements from Twitch's chief product officer, who justified the automatic enrollment of all users in the live session on August 12 by saying, “If it was opt-in, nobody would opt in. That’s honestly the answer.” When queried about whether any user content had already contributed to AI training, he replied, “I don’t actually know the answer to that question because I don’t know what Amazon has done in terms of model training and what they’ve used and not used.”

      In 2024, while serving as chief monetization officer, Minton acknowledged at an event that Amazon utilizes Twitch to train AI models, stating, “Yeah, for sure,” and specified it was for prototyping rather than any production-level application.

      Functionality of the setting

      The opt-out setting applies to channels, not individual users. The complaint cites Twitch's help pages stating: “If you chat on someone else’s stream, their opt-out preferences govern if that chat can be used for training.” The complaint asserts that Twitch provided no prior notice of the change, lacking email communication or pop-up alerts. A reporter, Zach Bussey, discovered the setting, and users have reported that the toggle reverts to enabled after being turned off. Coverage of the opt-out setting occurred on August 13.

      The four claims made

      Pandiscia claims breach of implied contract, unjust enrichment, breach of express contract, and violation of California's Unfair Competition Law under Business and Professions Code section 17200. The implied contract claim names only Twitch, while the other three include both companies.

      The filing does not contain a copyright claim, explaining that Twitch streams and videos “are not registered with the US Copyright Office.” It asserts that this lack of registration “does not render them valueless or leave them unprotected.” The California Invasion of Privacy Act is mentioned only as background for the unfair competition claim and