A Class Action Over Twitch Streams
A streamer has taken a stand against Twitch and Amazon using his content to train AI models. Warren Pandiscia, a streamer based in Connecticut, filed a class action lawsuit earlier this week. The complaint accuses Twitch and Amazon of failing to obtain licensing or permission before using his streams and videos as a “dataset necessary to fuel Amazon’s AI products.”
Key Takeaways
- The 37-page complaint was filed in the US District Court for the Northern District of California and demands a jury trial, asserting four causes of action including breach of implied contract, breach of express contract, unjust enrichment and violation of California’s Unfair Competition Law.
- Twitch confirmed the practice on August 12 with a post adding a setting that lets creators opt out of having channel content used to train generative AI models across Amazon, with every channel enrolled by default.
- The complaint alleges the data collection began as early as 2024, well before Twitch disclosed it publicly or offered any way to decline.
Pandiscia has streamed for roughly a decade and has more than 900 followers. The filing says he had no reason to believe Twitch and Amazon would use his content to train generative AI products, that he remains opposed to that use, and that he switched the setting off as soon as he learned it existed.
“Because Amazon AI products are commercialized, Amazon had an overwhelming incentive to acquire training data on an unprecedented scale. Rather than negotiate for lawful licenses or seek permission, defendants accessed the Twitch streams and videos to utilize them as a massive dataset necessary to fuel Amazon’s AI products,” Pandiscia said in the complaint.
The complaint describes millions of livestreams, recorded videos, clips and chat logs copied without permission from every party captured in them. It argues that consent of that kind is structurally impossible to obtain, and that once material has trained a model, creators cannot pull it back out.
The Setting That Started It
Twitch’s announcement did not land well. Shortly after the post, the company hosted an episode of its recurring Patch Notes stream where a viewer asked chief product officer Mike Minton why the AI training setting was not opt-in.
“if it was opt-in, nobody would opt in,” Minton said.
That remark now appears in the litigation. The complaint criticises both the timing of the disclosure and the design of the control, noting that the setting attaches to individual channels rather than to accounts, and that it does not apply retroactively. A creator who opts out can still be captured on somebody else’s stream.
The suit also accuses the two companies of profiting from streamers who, it says, “lost money or property as a result of defendants’ unlawful, unfair and fraudulent practices,” and of treating creators as free training stock for separate commercial AI products.
Not the First Case of Its Kind
Amazon is not alone in facing this argument. Earlier this year three YouTube channels filed a class action against Apple, accusing the company of scraping copyrighted content to train its AI models. That case turns on a different legal theory, alleging Apple circumvented technical protections in violation of the Digital Millennium Copyright Act rather than breaching a contract with creators. Apple has moved to dismiss, arguing the videos were posted publicly and that anyone could watch them. The same plaintiffs have brought comparable suits against several other technology companies. The filing that first surfaced the Twitch case lays out the contract-based approach in detail.
The contractual framing may prove the more durable one. Copyright litigation over AI training has produced mixed and slow results, and one high-profile case ended without a clear answer on the central question, leaving the legal position unsettled. Suits built on platform terms of service and unfair competition law sidestep the fair-use debate entirely and ask a narrower question: what did the platform promise its users, and did it keep that promise?
Publishers have pursued the copyright route with mixed fortunes, including reference publishers suing over scraped articles. Meanwhile infrastructure providers have started building commercial alternatives, with one major network operator rolling out pay-per-use terms for AI crawlers so that content owners get paid when their material creates value. For individual creators, none of that helps much yet, which is why leverage in the creator economy keeps shifting toward whoever controls distribution. Neither Amazon nor Twitch has filed a public response to the complaint.
Swiss National Bank Official Says AI Could Push Inflation Higher Near Term
Robots Break Two Human Sprint Records at Beijing’s World Robot Games
Alibaba Releases Wan3.0 Video Model a Day After $10 Billion Share Sale
Faster Than Bolt, Beaten by a Cable: Robots Meet the Awkward Bits at Beijing World Humanoid Robot Games
Congressional Lawyers Spend Longer Fixing AI-Written Bills Than Writing Them
More news about Information processing
Russia is Reducing the Quality of its Aviation Fuels
Starcloud Raises $250 Million at $2.3 Billion Valuation for Building Orbital Data Centers
China Begins Its Largest Vehicle Recall Over Door Handle Safety, Tesla Also Affected
Uber Hit With Second-Largest GDPR Fine for Algorithmic Deactivations- Xiaomi Unveils Xring O3 Processor, TSMC to Build It on 3nm
