Amazon × Twitch
On August 12, 2026, Twitch, a subsidiary of Amazon, implemented a new setting that allows its parent company to use creator content for training generative AI models. [51, 59] The setting was enabled by default for all users, requiring them to actively opt-out if they did not want their future content used. [59] The change prompted a class-action lawsuit filed on August 20, 2026, alleging the data collection constitutes illegal wiretapping. [51]
The asset
Content from user channels, including live broadcasts, VODs, clips, chat messages, images, and text.
The record
- Seller status
- Going concern
- Sector
- video live streaming service
- Modalities
- text, images, video, audio
- Jurisdiction
- US
- Personal data
- A setting allows creators to opt out of having their channel content used for training. The opt-out applies only to future training, not past collection.
What it means if you hold data
This event represents a significant shift in how major platforms secure AI training rights, moving from explicit licensing deals to leveraging terms of service changes with default opt-in settings. For data brokers, it highlights a new, lower-cost method of data acquisition that may face legal and community challenges.