Twitch Adds Opt-Out Setting for Amazon Generative AI Training, Enabled by Default
This move shifts the burden of consent onto individual streamers, a pattern increasingly common across social platforms, and follows a report from August 1 that Twitch and Amazon were planning to use streams for AI training despite internal opposition.
Key Facts
- Twitch added the "Training for Generative AI" setting on August 13, 2026, in the "Security and Privacy" section of user settings.
- The setting is enabled by default, requiring streamers to opt out to prevent their channel content from being used for Amazon's generative AI training.
- The opt-out covers streams, video on demand, clips, stream chat, and images and text on the channel.
- Opting out does not stop all AI or machine learning use on Twitch; features like AutoMod and automatic caption generation may still use data.
- Twitch Chief Product Officer Mike Minton said the opt-out method was chosen because an opt-in approach would result in no one permitting use.
Reporting from 4 sources: Automaton, Denfaminicogamer, GIGAZINE, Game Spark.
On August 13, 2026, Twitch introduced a new setting that allows streamers to opt out of having their channel content used to train Amazon's generative AI content models. The setting, labeled "Training for Generative AI," is located in the "Security and Privacy" section of user settings and is enabled by default. When turned off, it prevents streams, video on demand, clips, stream chat, and images and text on the channel from being used in future training of Amazon's models for generating text, audio, images, and video. If a user chats on another streamer's channel, that channel's opt-out setting applies to their messages. The opt-out does not cover all AI or machine learning uses on Twitch; data may still be used for features like AutoMod and automatic caption generation, which Twitch says do not retain or use stream content for training new content models. The change has drawn backlash from users, and Twitch's Chief Product Officer Mike Minton explained that the opt-out method was chosen because an opt-in approach would result in no one permitting use. The company has not clarified whether it previously used channel content for AI training without this setting.
Twitch's new setting appears in the "Security and Privacy" tab, below other privacy controls. The toggle is on by default, meaning streamers who do not want their content used for Amazon's generative AI training must actively turn it off. The scope includes streams, video on demand, clips, chat messages, and images and text posted on the channel.
Twitch's community lead Merry Kish and Chief Product Officer Mike Minton discussed the change in a stream that drew about 3,000 viewers, according to TechCrunch. Minton said the opt-out design was chosen because with an opt-in model, "no one would permit its use." Kish noted that other platforms like Meta also use content for AI training and framed the opt-out as a response to community concerns.
Streamer Zack Bussey had reported on August 1, citing multiple sources, that Twitch and Amazon planned to use streams for AI training and that an announcement was expected within months. Whether Twitch used content for training before this setting existed remains unclear. In 2024, Minton said at a The Information event that Twitch data was not used in production but was used for prototype development, suggesting training may have been happening for two years.
Synthesized by Yomimono from the 4 cited sources below, including Japanese-language reporting where cited, then editorially reviewed before publishing.