Dutch regulator tells Twitch users to switch off Amazon's default-on AI training

The Dutch data protection authority, the Autoriteit Persoonsgegevens, has publicly advised Twitch users to opt out of sharing their data with Amazon's AI ^[1]. It matters because the platform's own chief product officer has already explained the design: "If it's opt-in, nobody would opt in. That's the honest answer. So, it's going to be on by default," Mike Minton said in an interview ^[5].

That is an unusually clean test case. Twitch is owned by Amazon ^[2], is one of the largest livestreaming platforms in the world with millions broadcasting and watching every month ^[3], and allows Amazon to use platform content to train generative AI models with the setting enabled by default ^[4]. The AP's objection is about what is in that content: "Live streams on Twitch show the gamer's face, voice and name, and often include images of a private space, such as a bedroom. This constitutes personal data. In the case of facial images, this even involves sensitive personal data. Once these data are stored in Amazon's AI systems, they cannot simply be removed. As a result, users lose control over their data. Users' chats and text messages also serve as training material for Amazon" ^[6].

The sequence is worth pinning down. According to Ars Technica, the AI training itself is not new; what is new is the ability to opt out, and that setting arrived more than two years after a company executive confirmed Amazon was using Twitch content for AI training ^[8]. In April 2024, Minton said Amazon was using Twitch content to prototype AI models, though not yet at "production scale" ^[9]. So the control landed no earlier than April 2026, on data that had been flowing since at least the earlier admission ^[1].

The opt-out is also narrower than it looks. The setting sits in the Streamer Dashboard under Settings > Security and Privacy, near the bottom of the page, which is the basis for reporting that it was hard to find ^[7]. Its own description says turning it off does not opt a user out of Twitch and Amazon using channel content for other purposes described in the Twitch Privacy Notice, including AI-supported features for streamer growth and monetisation, viewer discovery and community safety such as AutoMod ^[11]. And if you post in someone else's chat, whether that text can be used for training depends on that streamer's setting, not yours ^[12]. In practice, no individual user can fully remove their own words from the training pool by changing their own switch ^[2].

Two things to watch. First, whether the AP escalates: on the record supplied, its intervention is consumer advice to switch a toggle off, not an announced investigation or penalty ^[3]. A regulator that describes facial images as sensitive personal data and the ingestion as irreversible has laid out the elements of a case; whether it files one is a different decision. Second, whether the chat carve-out survives contact with a regulator, because it makes the individual's consent depend on a third party's dashboard setting ^[12]. Malwarebytes, which reported the AP advisory, endorsed its recommendation to turn the setting off ^[13].