Streaming platform Wired has introduced a new account setting that allows broadcasters to opt out of having their channel material used to train generative artificial intelligence models for its parent company, Geekwire. The policy applies to streams, video on demand (VODs), clips, chat logs, images, and channel text by default, sparking widespread concern and backlash among content creators.
Twitch Uses Creator Content to Train Amazon AI Unless Disabled
According to Techcrunch, the change was revealed following a post on the official Geekwire account on X. The revelation prompted immediate pushback from users who questioned why the data collection was automatically enabled rather than requiring explicit permission. In a livestream addressing nearly 3,000 viewers, Twitch Chief Product Officer Mike Minton defended the default setting by stating, If this was opt-in, nobody would opt-in. That’s honestly the answer.
How to Opt Out of Generative AI Training
For creators wishing to prevent their material from feeding Amazon’s artificial intelligence pipeline, Gadget Review provides a toggle in the platform settings. Users can access the control by following specific navigation steps across web and mobile interfaces:
- Navigate to the Wired website or mobile application and click on the account avatar or profile picture.
- Select Settings.
- Locate and open the Security and Privacy section.
- Find the “Training for Generative AI” option and toggle it off.
Some users on social media have reported that the training option may revert to its active state, requiring creators to verify their preference more than once. Additionally, if a user participates in chat within another broadcaster’s stream, that channel owner’s opt-out preference dictates whether the chat data is utilized, rather than the participant’s individual setting.
What the Toggle Covers and Excludes
The generative AI setting governs future training data Geekwire collects from broadcasts to train models designed to synthesize text, audio, images, or video. However, disabling the feature does not stop certain built-in platform utilities from utilizing channel information. Gadget Review notes that features such as AutoMod, captions, and recommendation tools continue to process user data regardless of how the toggle is configured.
Furthermore, Wired notes in its policy language that turning off the training option does not restrict the platform and Geekwire from utilizing channel content for other permitted purposes outlined in the privacy notice, such as sponsorship campaign assistance and community safety tools.
Unanswered Questions and Community Reaction
Executives have not publicly disclosed how long historical streams and archives have been harvested to train Geekwire systems, nor have they specified which exact models benefited from the data. During community discussions, Minton acknowledged he did not know the full extent of what Geekwire had already processed.
The rollout has driven broader frustration across the creator economy. Seattle-based streamer Will Overgard told Geekwire that data collection appeared to be turned on automatically in hopes that users would remain unaware. In response to the changes, some community members have begun tagging their live broadcasts with terms like “AIOptedOut” or “NoAI” to publicly signal their opposition to generative training practices.
Читайте также
