Amazon will use content from Twitch livestreams to train artificial intelligence models by default, prompting widespread criticism from the platform's streaming community. Users must actively opt out to prevent their content from being used in AI training, a reversal of standard opt-in practices common elsewhere in the technology industry.
The policy applies to video, audio, and chat content on Twitch, which Amazon acquired in 2014 for $970 million. Streamers discovered the change through updated terms of service that take effect immediately. The company characterized the practice as ongoing, suggesting Twitch content has already been used to train AI systems for an unspecified period before the disclosure.
Streamers expressed anger over the default opt-in approach, arguing they should have explicit control over how their creative work is used. Many creators rely on Twitch for income through subscriptions, donations, and advertisements, and view their streams as proprietary content. The backlash reflects broader tensions in the creator economy as technology companies seek training data for AI development while content producers demand compensation and control.
Twitch provided instructions for users who wish to opt out of AI training, though the process requires navigating account settings. The platform has not disclosed which specific AI products use Twitch data or whether Amazon's broader suite of AI services, including its cloud computing division's machine learning tools, benefit from the training material.
The controversy adds to growing scrutiny of how technology companies source training data for AI systems. Publishers, artists, and other content creators have increasingly challenged the practice of using copyrighted or user-generated material without explicit permission or payment. Amazon has not announced changes to the policy in response to the criticism.