Streamers Are Suing Twitch and Amazon — and the Data Can Never Be Unlearned

In the first few months of 2026 alone, streamers on Twitch produced more than 215 million hours of content. For Amazon, that figure represented not just entertainment — but training data. A class-action lawsuit filed on 20 August in the U.S. District Court for the Northern District of California accuses Amazon and its streaming platform Twitch of scraping creators’ videos, audio, and clips to train generative AI models — without permission, without compensation, and before anyone was told they could say no.

What Twitch Quietly Did in 2024

The lawsuit, filed by Connecticut-based streamer Warren Pandiscia, alleges that Amazon began collecting Twitch users’ content for AI training as early as 2024. That timeline is significant: the platform only introduced an opt-out setting in August 2026, meaning the scraping allegedly began two years before streamers were given any mechanism to stop it. Mike Minton, then Twitch’s chief monetisation officer, reportedly acknowledged in 2024 that content was already being used for AI development in a “prototyping” capacity. He’s now the platform’s chief product officer — and became the face of the backlash when the opt-out setting launched. Pandiscia holds just over 900 followers. He says he opted out the moment he found out the setting existed, and argues he would have done so far earlier had the practice been disclosed.

The Opt-Out That Isn’t Much of an Opt-Out

The setting, buried under Security and Privacy in the Streamer Dashboard, reads: “Allow your channel content to train generative AI content models at Amazon.” It is switched on by default. The complaint identifies a significant loophole: settings apply per channel, not per user. A streamer who has opted out can still have their content captured if they appear, chat, or are clipped on a channel that has not opted out. In practice, it is nearly impossible to fully escape the net. When pressed during a Patch Notes stream, Minton did not apologise for the design.

“If it was opt-in, nobody would opt in. That’s honestly the answer.” — Mike Minton, Twitch Chief Product Officer

Why Creators Say the Harm Is Permanent

The lawsuit’s most striking argument isn’t about money — it’s about the irreversibility of the alleged harm. “Content creators such as plaintiff and the class members will never be able to claw back the intellectual property unlawfully copied and used by defendants to train Amazon’s generative AI,” the 37-page complaint states. “Once AI ingests content, that content is stored in its neural network, and not capable of deletion or retraction.” Twitch’s own FAQ acknowledges broad use of creator content: audio may be used to “refine models that create speech to text” to improve automatic captions both on Twitch and across Amazon. The company argues this falls within its terms of service and helps the streaming experience. Pandiscia disputes that framing. His complaint argues the companies made implicit commitments about how creators’ content would be treated, and that using it as “free training stock for separate commercial AI products” breaks those commitments. The suit lists breach of implied and express contract, unjust enrichment, and unfair business practices under California law.

What the Lawsuit Is Seeking

The claims are broad. Pandiscia is seeking injunctive relief — an order preventing Amazon from continuing the alleged practice — alongside damages, restitution, and disgorgement of profits. No specific monetary amount is named in the complaint, which is standard at the class action filing stage. The class, if certified, would encompass millions of streamers. Amazon, which now owns EA after a $55 billion acquisition closed earlier this month, has not commented on the lawsuit. Neither has Twitch. The case raises a question the entire creator economy is circling: does a default-on toggle, buried four layers deep in settings, constitute meaningful consent?