Twitch users outraged as Amazon uses their content to train AI in opt-out feature
- Published
Twitch has been criticised by users after it emerged the popular streaming platform allowed their content to be used to train Amazon's AI models.
Amazon owns Twitch and a setting allowing the US tech giant to take data generated by creators and audiences to train AI is turned on by default.
Users can opt out of that, Twitch said on Wednesday, external, but the announcement sparked a backlash with some users questioning why it was allowed in the first place.
Twitch's chief product officer Mike Minton, said it was "respecting" users by letting them opt out but on why data was collected by default, he admitted: "If it's opt-in, nobody would opt-in. That's the honest answer."
What is Amazon using the content for?
Users who remain opted in could have any of their channel content used for training, Minton said during a livestream, adding that the data collected will not be re-sold to other companies.
In Twitch's own FAQs about its use of AI, external, it says if users do not opt out, their content could be used to train generative AI models.
Generative AI is a type of artificial intelligence which creates new content, such as text, images and video. Chabots like OpenAI's ChatGPT and Google's Gemini are both examples of the tech.
For example, Twitch said a person's audio might be used to "refine models that create speech to text". It said this would help improve automatic subtitles on Twitch streams as well as Amazon videos.
Some also questioned, external what the implications could be for game developers, given the feature would also seemingly train Amazon's AI models on the countless video games being played by streamers.
"On by default is criminal.... the AI narrative push is so draining," said one streamer, external underneath Twitch Support's post about the feature.
The BBC has contacted Twitch and Amazon for comment.
How to opt-out
On the same livestream, Mary Kish, head of community at Twitch, took viewers through a tutorial on how to disable generative AI training through a user's channel settings.
Those that do want to disable it would need to navigate to the Settings tab in their Streamer Dashboard, click the Security and Privacy tab, then scroll down to near the bottom of the options and toggle off "training for Generative AI".
Doing so will prevent Amazon from using streams, clips, images, chats and other channel content to train its generative AI models.
Some users have since said despite toggling off the option, they have then returned to the settings to find the feature turned on again.
"Went in and toggled the switch off, exited, went back in, switch is still on," one said, external.
Data collection for AI training has become an industry standard, Kish said.
"We don't expect you to be happy or excited about this. I don't expect anyone to react to this favourably," she added.
The stream's chat column was flooded with hundreds of comments from users criticising the move.
One viewer wrote: "Nobody would opt in because nobody wants to feed AI with our creativity and content."
Another said Twitch had an "opportunity to set an industry standard" to push back against data collection for AI.
It is not clear when Amazon began collecting Twitch users' data to train its AI models.
Minton said during the stream that he did not know if users' data had already been scraped for training, adding that he was unsure what Amazon "has done in terms of model training and what they've used and not used."
Amazon, which bought Twitch for nearly $1bn (£740m) in 2014, runs a host of AI services and has heavily invested in the its development as it seeks to compete with Google, Meta and other technology giants.
Facts Only
* Twitch allows content to be used to train Amazon's AI models by default.
* Amazon owns Twitch.
* A setting allowing data training is enabled by default.
* Users can opt out of this data usage.
* If users do not opt out, their content could be used to train generative AI models.
* Audio may be used to refine speech-to-text models for automatic subtitles on Twitch streams and Amazon videos.
* The data collected will not be re-sold to other companies.
* A user tutorial exists to disable generative AI training in channel settings.
* Some users reported that disabling the setting did not permanently stop the training.
* Amazon acquired Twitch in 2014.
Executive Summary
Full Take
The situation reveals a tension between default system design and user expectation regarding data privacy, especially when powerful entities control both platforms. The concept of "opt-out" is presented as respect, yet the default setting implies a prioritization of corporate utility over immediate user autonomy, suggesting that permission is framed as an afterthought rather than a foundation for collection. The dynamic where users must actively engage to reclaim default privacy settings highlights a structural imbalance where inertia favors data harvesting. Furthermore, the skepticism expressed by users—that no one would opt in if it were optional—points toward a systemic lack of trust regarding the value exchange for this type of content. This pattern suggests that transparency about 'what has been done' (like potential scraping) is secondary to establishing a framework where consent is meaningful. The core implication centers on agency: whether a default setting reflecting mass data collection constitutes an abdication of responsibility, and how industry norms can be established to reverse the trend of AI training becoming an ambient feature rather than an explicit choice.
BRIDGE QUESTIONS:
What verifiable mechanisms exist to audit the extent of Amazon's utilization of Twitch content for AI training beyond the stated opt-out option? What governance structures could compel platforms to default to privacy settings by design, rather than requiring constant user intervention? How should the industry define the acceptable threshold for data use when leveraging creator and audience content for training generative models?
Sentinel — Human
The text reads like standard news reporting on a platform controversy, characterized by direct quotes and procedural details, suggesting a human journalistic origin.