Twitch has confirmed that content from streams can now be used to train generative AI models across Amazon, and the feature is enabled by default for streamers.

Rumors about Twitch introducing an AI training program had been circulating for several weeks, but the platform indirectly confirmed the system through a Twitch Support announcement about a new setting that allows creators to opt out.

That means streamers are automatically enrolled unless they manually change their settings, a decision that has already triggered significant backlash from the Twitch community. Responses to the announcement have largely focused on showing other users how to disable the feature, alongside plenty of criticism directed at the platform.

The reaction has also extended to Twitch's UserVoice forum. A thread calling for all AI features to be optional and opt-out by default has accumulated nearly 14,000 votes and hundreds of pages of responses, showing just how strongly some creators feel about the issue.

Twitch Chief Product Officer Mike Minton addressed the decision during a livestream and gave a remarkably direct explanation for why the feature is enabled by default: "If it was opt-in, nobody would opt in."

Opting out is relatively straightforward. Streamers can open the Security and Privacy section of their Twitch settings and scroll down to the "Training for Generative AI" option. From there, they can disable the feature.

The bigger problem is whether streamers know the option exists in the first place.

Twitch did not announce the system through a prominent post on its official blog. Instead, the information was shared through a Twitch Support post about the new setting. That has raised concerns that many creators could remain enrolled simply because they never realized they had the option to opt out.

The situation became even more confusing when Twitch responded to a post about the feature by saying, "Twitch is not training GenAI models on streamer content."

That statement appears difficult to reconcile with Twitch's own account settings FAQ, which explicitly discusses the use of streamer content to train generative AI models.

The FAQ explains that generative AI content models can create or synthesize things such as text and images after receiving instructions. It also states that allowing training means a user's content may be used for future generative AI model improvements.

Twitch provides audio as an example, explaining that stream audio could help improve models that generate speech-to-text. Those improvements could benefit Twitch's captioning systems, but could also be used to improve captioning throughout Amazon.

That raises an obvious question about Twitch's statement that it is not training generative AI models on streamer content. One possible interpretation is that Twitch itself is not training the models, but Amazon is using the content to do so. Since Twitch is owned by Amazon, however, the distinction may not mean much to the creators whose content is being used.

Twitch has been contacted for clarification on the apparent contradiction.

There is another important detail in the platform's FAQ. Opting out of generative AI training does not prevent Twitch from using AI and machine learning for other purposes. Twitch can still use streams, VODs, and other data for AI-supported features such as automatically generated captions.

The distinction is that this data is not retained for training models designed to generate new content.

The controversy also raises questions that extend beyond individual streamers. Game developers could potentially be affected when creators stream their games while participating in the AI training program.

Mike Futter, co-founder and analyst at game consultancy F-Squared, argued that the implications could be particularly serious for developers that do not want their creative work used for AI training.

"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 also warned that studios could soon begin consulting their lawyers over the issue.

"I think a lot of studios are going to be talking to their lawyers over the next couple of days," Futter told PC Gamer. "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.

"In a period where our industry needs as much support as possible, especially from the companies that use our content every day to enrich themselves, this reads as extremely hostile."

Futter, who also serves as director of operations and publishing at indie developer Causeway Studios, added that the studio is strongly opposed to AI and sees Twitch's new system as a potential threat to its creative work.

"We're a staunchly anti-AI studio, and this now creates a real threat to our creative work," he said.

While Twitch has faced its share of controversies and policy changes over the years, the generative AI training program could prove particularly difficult for the platform to navigate. The issue affects not only how Twitch treats its creators, but potentially how their content interacts with the intellectual property of the developers and publishers whose games they stream.

Futter believes that pushing back against Twitch and Amazon will require a large number of studios and creators to speak out together.

"This is David and Goliath. We're going to need a critical mass of studios and creators to be loud about this."

Twitch has reversed or changed course on controversial decisions before, so it is possible that the current backlash could eventually force the company to rethink its approach. However, Amazon has been heavily investing in generative AI, making a complete retreat from the technology seem unlikely.

For now, Twitch's new AI training system has created a complicated situation for the platform. The feature is enabled by default, the communication around it has been unclear, and creators are now being asked to decide whether they are comfortable allowing their content to contribute to Amazon's generative AI efforts.

Whatever happens next, Twitch is likely to face considerably more questions about exactly how streamer content is being used and who ultimately benefits from it.