Twitch is facing backlash from users after it emerged that content generated by creators and audiences on the streaming platform could be used to train Amazon’s artificial intelligence (AI) models.
The Amazon-owned platform had an AI training setting turned on by default, allowing eligible channel content to be used for generative AI training. Twitch has now introduced an option for users to opt out of the programme.
Twitch AI Training Setting Sparks User Backlash
Twitch Chief Product Officer Mike Minton acknowledged the criticism during a livestream, saying the company is trying to respect users by allowing them to disable AI training. However, he also admitted that making the feature opt-in would likely have resulted in very few users agreeing to it.
“If it’s opt-in, nobody would opt-in. That’s the honest answer,” Minton said during the livestream.
For users who remain opted in, Twitch said content from their channels could be used for AI training. This includes streams, clips, images, chats and other channel content. Twitch also said the data collected for this purpose will not be resold to other companies.
Twitch Head of Community Mary Kish also demonstrated how users can disable generative AI training through their channel settings. Turning off the option prevents Amazon from using their Twitch content to train its generative AI models.
The move quickly triggered criticism in Twitch’s livestream chat, with hundreds of users questioning why AI training was enabled by default rather than requiring creators to give explicit consent.
One viewer argued that users did not want their creativity and content to be used to train AI models. Another said Twitch had an opportunity to establish a different approach to AI data collection across the industry.
Twitch Says AI Data Collection Is Becoming Industry Standard
Kish defended the broader context of the move by saying that data collection for AI training has become an industry standard. At the same time, she acknowledged that users were unlikely to welcome the announcement.
“We don’t expect you to be happy or excited about this,” Kish said during the livestream, adding that she did not expect users to react favourably.
The company has also not clarified when Amazon first began collecting Twitch data for AI training. Minton said he did not know whether users’ content had already been scraped or used to train models, or exactly what Amazon had done with the data.
The uncertainty around past data use has added another layer to the backlash, particularly among creators who want greater control over how their content is used to develop AI systems.
Amazon acquired Twitch for nearly $1 billion in 2014 and has since expanded its investments in artificial intelligence as it competes with other major technology companies, including Google and Meta.
Amazon has been contacted for comment on Twitch’s AI training policy and whether creator data had already been used to train its AI models.
FAQs
- What is Twitch AI training?
Twitch AI training refers to the use of eligible Twitch content, such as streams, clips, images and chats, to train generative AI models.
- Does Twitch use creator content to train AI?
Yes. Twitch says eligible creator content can be used for AI training when the AI training setting is enabled.
- Is Twitch AI training enabled by default?
Yes. The AI training setting was enabled by default for eligible channels, which has led to criticism from Twitch users and creators.
- Can Twitch creators opt out of AI training?
Yes. Twitch has introduced an option that allows creators to disable the AI training setting through their channel settings.
- How do I opt out of Twitch AI training?
Creators can turn off the generative AI training option through their Twitch channel settings to prevent eligible content from being used for AI training.
