Twitch’s latest policy change gives creators and viewers the power to prevent Amazon from pulling streams, clips, chats and other channel data for AI training. The default was to allow Amazon to use that content – a practice that took the gaming community to the brink of revolt.
After Amazon reported that it had been harvesting Twitch data for years, partnership details surfaced and the battle over data rights rattled the platform. In response, Twitch’s team told users that opting out is an irreversible choice unless they later re‑enable the setting.
Amazon’s growing investment in generative AI – competing with Google, Meta and other giants – has prompted scrutiny over how it acquires training data. Twitch clarified that data harvested under the new option will be used solely for Amazon’s own AI models and will not be resold.
During a live stream, Twitch chief product officer Mike Minton admitted that a default ‘opt‑in’ works because “if it’s opt‑in, nobody would opt‑in.” He emphasized that the company respects user wishes but acknowledges the persistent industry trend of data collection for AI usage.
The community’s reaction was instant and largely negative, with comments such as “Nobody would opt in because nobody wants to feed AI with our creativity.” Mary Kish walked users through disabling generative AI training in channel settings but also warned that the move was “not likely to get a positive reaction.”
It remains unclear when Amazon began gathering Twitch data and how much has already been scraped. Minton acknowledged uncertainty, saying he did not know whether data had already been collected for training purposes.
The decision marks a significant power shift for gamers concerned over data usage, and it will likely influence how other streaming platforms negotiate data access agreements. The broader debate continues over the ethics of using user-generated content to fuel powerful AI systems without explicit consent.
