Twitch v30.7 enables AI training on user videos by default in August 2026, sparks backlash

The update arrived alongside a major policy shift from Twitch's parent company, Amazon. A new account setting, discovered around August 12, allows the company to use channel content—including streams, VODs, clips, and chats—to train its generative AI models.^[2] This setting was enabled by default for all users, a decision that drew immediate and widespread criticism. Twitch's own support documentation confirms the setting is for "future Gen AI model improvements" across Amazon.

While official release notes for version 30.7 were not published, the timing of the app update and the policy discovery has linked the two events in the public eye, fueling negative sentiment.

Compounding the policy controversy, version 30.7 introduced a significant software defect. Users overwhelmingly report that the 'followed channels' page is now broken. When attempting to scroll down the list, the page automatically refreshes and snaps back to the top, making it impossible to browse followed streamers beyond the initial view.

This "snap back" behavior renders a core navigation feature unusable. As one 1-star reviewer on v30.7 described it: "This new update made scrolling impossible on my followers page everytime I scroll down it immediately snaps me back to the top of the page it’s so frustrating."

User reception to version 30.7 has been intensely negative, causing the app's average ratingGlossaryStar RatingThe 1-5 star average users give your app on the App Store and Google Play — a primary ranking signal and one of the biggest conversion drivers on your product page. to fall from 2.75 to 1.55 stars. The combination of the AI policy and the scrolling bug dominates new reviews. A 1-star review for v30.7 states, "New ai training — Wonderful app, amazing people, fun streamers but recently they forced every streamer into allowing ai to help train new streamers, you can opt out of it but the default is on, why Twitch, why?" Another user on the same version wrote simply: "They enabled AI Training on videos :/ ://// Stop being scummy like everyone else and earn a personality."

The decision to make AI data collection opt-out rather than opt-in appears to be a deliberate strategy. During a livestream, Twitch's Chief Product Officer, Mike Minton, provided a candid explanation for the choice.^[3] He was quoted by Forbes and other outlets as stating, "If it was opt-in, nobody would opt-in." This admission suggests the company anticipated low participation and chose a default-on setting to maximize its data acquisition for AI development, accepting the risk of community backlash as a trade-off.

Twitch now faces significant pressure to address both the controversial policy and the critical bug affecting a user base of over 2.8 million monthly downloads on iOS alone. A patch for the scrolling defect is likely imminent to restore core functionality. However, the long-term damage to creator trust appears contingent on how the company handles the AI data policy. The incident highlights a growing tension between platform governance and the data acquisition goals of parent companies like Amazon. Twitch's next move will likely signal its priorities.