Twitch and Amazon hit with lawsuit for training AI with streamers’ content
A Connecticut-based streamer, Warren Pandiscia, has filed a class action lawsuit against Twitch and its parent company Amazon, alleging that his streams and videos were used without consent to train Amazon's AI models. The complaint accuses the companies of breaching contract by treating streamers' content as "free training stock" for commercial AI products, without obtaining licensing or permission, and claims streamers suffered financial harm as a result.
The lawsuit follows Twitch's introduction earlier this month of an opt-out setting allowing streamers to prevent their channel content being used for generative AI training across Amazon, a system enabled by default rather than requiring opt-in. Twitch's chief product officer, Mike Minton, reportedly justified the opt-out approach by saying "if it was opt-in, nobody would opt in." This is not an isolated case, as three YouTubers separately sued Apple earlier this year over similar allegations of scraping copyrighted content to train AI models.
- Streamer sues Twitch and Amazon over unconsented AI training use
- Twitch's AI data-sharing setting is opt-out, not opt-in, by default
- Follows similar YouTuber lawsuit against Apple over AI training data
New here? Start with this
Twitch is a live-streaming platform, mainly used by gamers, owned by Amazon. Like other big tech firms, Amazon has been building generative AI models, and streamers' broadcasts and archived videos are a huge pool of video and audio material that could potentially be used to train them.
The case has been brought by Warren Pandiscia, a streamer from Connecticut, on behalf of a wider group of streamers, arguing that their content was used to train Amazon's AI without their permission or any payment, breaching the agreements they had with Twitch. It matters because it raises a question affecting many online creators: whether platforms can use material people upload for one purpose to build commercial AI products, and who should have a say or benefit when that happens.
This is part of a broader pattern of similar legal challenges, including a separate case brought by YouTubers against Apple, as courts and companies work out the rules around AI firms using creators' content to train their systems.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Streamers argue that consent must be meaningfully sought, not buried in a default setting few users will notice or understand, and that broadcasting a stream publicly is not the same as licensing it for commercial AI training. They contend that their performances have real commercial value, built through years of effort and audience-building, and that platforms should not be permitted to repurpose that work into training data for profit-generating products without explicit permission or compensation. The opt-out design, they say, reveals an awareness that creators would object if properly asked, making the arrangement feel less like informed consent and more like exploitation of inertia.
The case against
Twitch and Amazon can reasonably argue that their terms of service, which users agree to when streaming, already grant broad licences to use content for platform improvement and development, and that offering any opt-out at all goes beyond what many digital platforms provide. They may contend that training AI on publicly broadcast material is a transformative use that benefits the wider streaming ecosystem, funding new features and tools rather than replacing creators' work outright. An opt-in requirement, they might argue, would be commercially unworkable for a feature still being tested and refined, and offering a clear, accessible opt-out strikes a fair balance between innovation and creator control.