Community opposition has also surfaced through Twitch's feedback system. A UserVoice proposal calling for AI features to be optional and disabled by default had accumulated nearly 14,000 votes and hundreds of pages of discussion. The way Twitch disclosed the change has added to the criticism, with the company framing its announcement around the availability of an opt-out setting rather than leading with Amazon's ability to use creator content for AI training.
Twitch describes the program as extending beyond products operated directly by the streaming service. Its account settings FAQ says creator content may contribute to future improvements of generative AI models, including systems used elsewhere within Amazon. The company gives audio processing as one example, saying creator audio could help improve speech-to-text technology used for captions on Twitch as well as similar capabilities across Amazon.
That explanation created some confusion after Twitch Support wrote on X that “Twitch is not training GenAI models on streamer content,” while also explaining that its new control lets users decline Amazon's generative AI training. Twitch's FAQ separately states that content can be used for future model improvements when creators permit training. Asked during the livestream whether creator videos had already been used by Amazon before the setting appeared, Minton said he did not know. “I don’t actually know the answer to that question because I don’t know what Amazon has done in terms of model training and what they’ve used and not used,” he said.
Turning off the setting does not prevent every form of artificial intelligence or machine learning processing on Twitch. According to the platform, creator data can still support AI-assisted functions such as automatically generated captions. Twitch distinguishes those uses from retaining content to train models capable of generating new material. Kish said providing a way to disable generative AI training was “a reflection of reacting to this community’s voice that you are not wanting to train Gen AI models, and we want to give you that option.”
The scope of Twitch content also raises questions beyond streamers themselves because livestreams frequently contain games and other material created by third parties. Mike Futter, co-founder and analyst at consultancy F-Squared and director of operations and publishing at indie developer Causeway Studios, questioned what happens when a participating streamer broadcasts a developer's game. “What happens if a creator streams your game and is opted in? Does Twitch think it has the right to feed a developer's game into its AI slop factory?” Futter said.
Futter argued that the policy could force studios to consider the implications for their intellectual property when streamers broadcast their games. “I think a lot of studios are going to be talking to their lawyers over the next couple of days,” he said. “For those of us who have avoided AI like the plague, this is a clear and present danger to our businesses and our creative work.”
Creators can disable generative AI training through their account settings, so participation is not irreversible. The dispute instead centers on Twitch's decision to make participation the default, a choice Minton openly acknowledged would bring more creators into the program than a voluntary system. That default has become the core of the backlash as Twitch gives Amazon access to a potentially large pool of creator content for improving generative AI models.
This analysis is based on reporting from PC Gamer.
Image courtesy of Nick Barclay / The Verge.
This article was generated with AI assistance and reviewed for accuracy and quality.
