Twitch Streamer Sues Amazon, Alleging AI Training on Creator Content Without Consent

A Twitch streamer with roughly 1,000 followers has filed a proposed class action against Twitch and its parent company Amazon, alleging that the platform fed creators' broadcasts into generative AI training without their consent or compensation.

The plaintiff, Warren Pandiscia, filed a 37-page complaint in the Northern District of California on 23 August 2026. The suit accuses Amazon and Twitch Interactive of illegally scraping millions of videos from Twitch's digital video library to train AI models without permission, breaching creators' implied agreements with the platform and violating state unfair competition law (Eurogamer).

At the centre of the dispute is a disclosure Twitch made on 12 August, when a post on X confirmed that the platform uses creator content — including live streams, past clips, chat logs and photos — to train Amazon's AI models. Crucially, the opt-out setting Twitch offered creators is not retroactive, meaning content already harvested before a streamer toggled the switch remains in the training data (Eurogamer).

Pandiscia claims he opted out as soon as the setting became available, but argues that Twitch and Amazon had been quietly using streamer content for AI training as far back as 2024 — before any disclosure or opt-out mechanism existed (Eurogamer). According to a report by Malwarebytes, Twitch added the opt-out option two years after it confirmed that such training had begun (Malwarebytes).

The opt-out's design has drawn further criticism. The setting is tied to individual channels rather than accounts, so a streamer who opts out of having their own content used can still appear in another channel's stream that has not opted out — and that footage remains fair game for AI training (Eurogamer).

Twitch chief product officer Mike Minton offered a blunt explanation for why the opt-out was switched on by default rather than left to creators to enable: "if it was opt in, nobody would opt in" (Eurogamer). The platform faced a wave of user criticism after the AI training practice came to light, with the BBC reporting on the backlash on 13 August (BBC News).

Amazon acquired Twitch in 2014 for roughly US$970 million. The platform has since grown into one of the largest live-streaming services in the world, making its archive of creator footage a substantial data pool — the kind that generative AI models, systems that learn patterns from large datasets to produce text, images or video, rely on to improve their output (Eurogamer).

For Twitch's millions of creators, the lawsuit raises a straightforward question: who owns the footage they broadcast, and what happens to it once it leaves their channel? The complaint seeks to represent a class of affected streamers, though the court has not yet certified the class. Neither Twitch nor Amazon has publicly commented on the litigation.