Amazon and Twitch Hit With Class-Action Lawsuit Over Using Streamers' Videos to Train AI

Key points

• A class-action lawsuit filed in 2026 accuses Amazon of using Twitch streamers' videos to train AI models without their consent or compensation.
• Twitch streamers produced more than 215 million hours of content in the first few months of 2026 alone.
• Connecticut-based streamer Warren Pandiscia brought the case on behalf of millions of Twitch users.
• Twitch's chief product officer admitted he did not know what user data Amazon had already used for AI training before an opt-out option was added.
• Amazon bought Twitch for nearly $1 billion in 2014 and has made AI a major focus since.

Imagine spending hundreds of hours building a streaming channel on Twitch, the live-video platform where millions of people broadcast gaming sessions, cooking shows and music performances every day. Now imagine finding out that every word you said, every clip you made, may have been quietly fed into an AI training system without your knowledge. That is what a new lawsuit claims happened to millions of Twitch creators.

Connecticut-based streamer Warren Pandiscia filed the class-action suit, a type of legal case where one person sues on behalf of a large group, against Amazon, which owns Twitch. The lawsuit alleges Amazon used creators' videos to train its AI models, the software systems that learn patterns from enormous amounts of data, without getting proper permission or paying the people whose work was used.

Amazon announced the practice in August and gave users a way to opt out. That timing is the core of the argument. Critics say a company should ask permission before using someone's work, not after.

There is a further wrinkle with the opt-out itself. Even if a streamer turns it off for their own channel, their voice and video can still be captured if they appear as a guest on someone else's stream where the setting remains switched on.

What exactly does Amazon say it is doing with the data?

Twitch's own FAQ states that a user's audio may be used to "refine models that create speech to text." The company says this helps improve automatic subtitles on Twitch streams and on Amazon's video service. Critics say that is still commercial use of someone else's creative work.

Twitch's chief product officer Mike Minton added fuel to the fire when he publicly admitted he did not know what Amazon "has done in terms of model training and what they've used and not used" before the opt-out was introduced. That statement is now central to the lawsuit's claim that creators were left in the dark.

What do the streamers want?

The lawsuit asks a court for two things: financial damages for creators whose content was already used, and a court order stopping Amazon from continuing the practice. Neither Amazon nor Twitch has commented publicly on the case.

The sheer scale here matters. Twitch users produced more than 215 million hours of content in the first few months of 2026 alone, as first reported by BBC Technology. That is an enormous library of free, human-generated material sitting inside a company now locked in a fierce AI race with Google, Meta and other tech giants.

How can streamers opt out right now?

If you stream on Twitch and want to turn off the AI training setting, go to your Streamer Dashboard, open the Security and Privacy tab, scroll to the bottom and toggle off "training for Generative AI." Remember: this only covers your own streams. It does not protect you when you appear on another creator's channel.

The broader question this case raises is one the whole industry is wrestling with. When a platform owns the pipes and the audience, who owns the performance?