Twitch AI Training Is On by Default: How to Opt Out
18:00 · August 13, 2026 · RSS APP - AI Security and Privacy

Twitch’s Amazon AI training setting is on by default. Here’s why creators are angry, how to opt out, and what the new control does not cover.
Summary
Twitch has added a privacy toggle that lets channel owners prevent their streams, VODs, clips, chat logs, images, and text from being used to train Amazon’s future generative models. The control sits under account-level Security and Privacy settings rather than the Creator Dashboard, and it is switched on by default for every account.
Because the toggle must be manually disabled, the change has drawn the same objections that followed Meta’s recent default-on policy for public Instagram content. Critics argue that meaningful consent requires an opt-in choice, while Twitch product leadership has stated that an opt-in approach would result in negligible participation. The setting applies only to prospective training runs and does not address any Twitch material Amazon may already have used. It also leaves untouched other AI-supported features on the platform that do not involve model training.
To disable the option, creators sign in, navigate to Settings > Security and Privacy, scroll to the Training for Generative AI section, and turn off the toggle that permits channel content to train Amazon’s generative models. The control therefore offers a narrow, forward-looking safeguard rather than a comprehensive block on Amazon’s use of Twitch data.
Why it matters
This article is highly relevant for privacy professionals as it highlights the ongoing tension between platform AI training data collection and user consent. It provides a practical example of opt-out defaults that professionals must monitor for GDPR compliance and data governance advising.






