The short version
- Twitch now allows users to disable the default setting that permits Amazon to use their streams and chat logs for AI model training.
- Executives stated that an opt-in system would result in negligible participation, justifying the current default configuration as necessary for data collection.
- It remains unclear when the data harvesting began or whether user content has already been incorporated into existing artificial intelligence models.
Twitch has introduced a mechanism allowing users to prevent Amazon from utilizing their livestream content for the training of generative artificial intelligence models. The announcement, made on Wednesday, confirmed that the streaming platform had been collecting data by default, a revelation that triggered significant criticism from its community. Users can now adjust their channel settings to opt out of this data sharing, which previously included streams, clips, images, chat logs, and other channel-related materials.
The decision follows intense backlash from creators who felt their creative work was being harvested without explicit consent. During a live session on the platform, Twitch executives addressed the controversy directly. Mike Minton, the chief product officer, explained that the default setting remains active because an opt-in alternative would likely result in almost no participation. He characterized this outcome as the honest reality of user behavior regarding data privacy settings.
Mary Kish, head of community at Twitch, guided viewers through the technical steps required to disable generative AI training features. She acknowledged that the company did not anticipate a positive reception to the news. Kish described the collection of data for AI purposes as an established industry norm, suggesting that while users might be unhappy with the practice, it is consistent with how many other content services operate in the current digital landscape.
The chat during the executive stream was filled with hundreds of comments expressing disapproval. Many users argued that they did not want their creativity and content to fuel artificial intelligence development. Some viewers criticized Twitch for missing an opportunity to establish a higher ethical standard within the industry by resisting such data collection practices. The sentiment reflected a broader unease among creators about how their digital footprint is being leveraged by large technology corporations.
Amazon, which acquired Twitch in 2014, has invested heavily in artificial intelligence development to compete with rivals like Google and Meta. The use of Twitch content supports these broader corporate goals. However, the specific timeline for when Amazon began scraping Twitch data for AI training remains ambiguous. Minton admitted during the stream that he did not know if user data had already been used for model training prior to this announcement.
Executives were unable to provide clear details on what specific data has been utilized or excluded from previous training cycles. This lack of transparency has added to the frustration among users who are concerned about the extent to which their past broadcasts may have already contributed to AI systems. The uncertainty regarding historical data usage complicates efforts for creators to fully understand the impact of the new opt-out feature.
While Twitch stated that collected data will not be resold to third-party companies, the primary concern remains the internal use of this information by Amazon. The platform’s top content creators, who command audiences in the tens of millions, are particularly affected by these changes. Their ability to control how their likeness and voice are used in emerging technologies is now a configurable setting rather than a guaranteed right.
The situation highlights the growing tension between user privacy expectations and the data-hungry nature of modern AI development. As more platforms adopt similar default settings, the burden shifts to individual users to actively manage their digital rights. Twitch’s move to allow opting out represents a concession to public pressure, even as executives maintain that the underlying practice is standard and necessary for business operations.
Looking ahead, it remains to be seen how many users will actually change their settings. The complexity of navigating privacy options often leads to low engagement, which aligns with Minton’s assertion that few would opt in if given the choice. For now, Twitch has provided the tools for resistance, but the default path continues to favor data collection for Amazon’s artificial intelligence initiatives.
This development underscores the evolving relationship between content creators and the platforms that host them. As AI becomes more integrated into digital services, questions about consent, compensation, and control over creative output are likely to become more frequent. Twitch’s handling of this issue may serve as a case study for other streaming and social media platforms facing similar pressures from their user bases.
Sources behind this briefing
Go to the original reporting
- BBC Business↗Twitch faces backlash over Amazon training AI on its users' livestreams
