Amazon is using content streamed on Twitch, its live gaming platform, to train generative AI models without explicit creator consent. Twitch streamers and the broader creator community have pushed back hard against the move, arguing that their broadcast footage should not feed AI development without permission or compensation.
The company's terms of service update allows Amazon to harvest channel content for machine learning purposes. Streamers discovered the change buried in updated policies, sparking immediate backlash across social media and streaming communities. Many creators view this as a violation of their intellectual property rights, particularly those who depend on Twitch revenue and build audiences around their original content.
This move reflects a broader industry trend. Tech companies from OpenAI to Google have faced lawsuits and regulatory scrutiny over unauthorized training data scraping. Twitch creators occupy a unique position: their content is already public and streamed live, yet many consider their performances and commentary proprietary creative work worthy of protection.
Amazon claims the data improves AI services across its ecosystem, including content recommendations and moderation tools. However, the lack of transparency and opt-in mechanisms troubles creators who see no direct benefit or revenue sharing. Some streamers have threatened to migrate to competing platforms like YouTube Gaming or kick.com.
The decision also exposes tensions between platform policies and creator expectations. Twitch built its user base on streamers trusting the platform with their livelihoods. Using that content for AI training without negotiated terms or compensation erodes that trust. Amazon faces pressure to either offer creators a choice in data usage or establish revenue-sharing models similar to those some platforms have tested with publishers and artists.