{"id":1660953,"date":"2026-08-12T22:17:08","date_gmt":"2026-08-12T22:17:08","guid":{"rendered":"http:\/\/UmqLY89bRf86QSL7SavnUb"},"modified":"2026-08-12T22:17:08","modified_gmt":"2026-08-12T22:17:08","slug":"twitch-under-fire-for-new-gen-ai-training-system-that-harvests-streamer-data-for-amazon-says-its-on-by-default-because-if-it-was-opt-in-nobody-would-opt-in","status":"publish","type":"post","link":"https:\/\/arcader.org\/news\/twitch-under-fire-for-new-gen-ai-training-system-that-harvests-streamer-data-for-amazon-says-its-on-by-default-because-if-it-was-opt-in-nobody-would-opt-in\/","title":{"rendered":"Twitch under fire for new gen AI training system that harvests streamer data for Amazon, says it’s on by default because ‘if it was opt-in, nobody would opt in’"},"content":{"rendered":"
\n

Twitch has revealed that it is now using content in streams “to train generative AI content models across Amazon,” and making matters worse, everyone is opted into the training program by default. <\/p>\n

Rumors of an AI training program have been kicking around for at least a few weeks now\u2014streamer Zach Bussey posted about it on Bluesky<\/a> back in July\u2014but it was confirmed today, very indirectly, by a Twitch Support announcement on X<\/a> <\/a>of a setting that lets streamers opt out.<\/p>\n

\n
\n
\n

\"We've<\/a><\/p>\n<\/div>\n<\/div>

(Image credit: Twitch (Twitter))<\/span><\/figcaption><\/figure>\n

That, naturally, means the default setting is opted-in, which has not gone over well: Responses to the message on X are predominantly either explanations of how to turn AI training off, or enthusiastic f-bombing. A Twitch User Voice<\/a> thread on the topic, originally posted in May, calling for all AI features to be “optional and opt-out by default,” has exploded to 228 pages, and nearly 14,000 votes in favor, as I write this.<\/p>\n

As for why the new generative AI training system is opt-in by default, Twitch chief product officer Mike Minton gave an unexpectedly honest answer during a livestream today, and by extension confirmed that Twitch is well aware of how people really feel about it: “If it was opt-in, nobody would opt in.” So, credit for at least being straight about it, I suppose.<\/p>\n

\n

they know we don’t want it. they do not care.<\/p>\n

\u2014 @witchytq.bsky.social (@witchytq.bsky.social.bsky.social<\/a>) 2026-08-12T22:17:07.747Z<\/a><\/p><\/blockquote>\n

It’s not particularly difficult to opt out of Twitch’s generative AI training scheme. Just go to the “security and privacy<\/a>” tab in your Twitch settings, then scroll down to the bottom where you’ll see the “Training for Generative AI” toggle. Click that (and I would strongly encourage you to do so), and you’re done.<\/p>\n

The problem is that ease of use does no good if people don’t know it’s there, and Twitch does not seem to be making great efforts to ensure everyone is aware of it. Revealing the new system through a post about a related feature on the Twitch Support channel rather than a conventional announcement on the Twitch Blog is hardly a high-visibility way to get the word out, and a response<\/a> to a more recent post by Bussey about the feature muddied the waters by saying “Twitch is not training GenAI models on streamer content.”<\/p>\n

\n
\n
\n

\"Just<\/a><\/p>\n<\/div>\n<\/div>

(Image credit: Twitch (Twitter))<\/span><\/figcaption><\/figure>\n

But Twitch’s own account settings FAQ<\/a> indicates that’s exactly what it’s doing. Under the question “What are ‘generative AI content models,’ and what does it mean for my content to be used to train them?” Twitch wrote:<\/p>\n

“Generative AI content models are AI tools that can help create or synthesize content outputs like text or images when given instructions or prompts because they are also trained on content. When you allow training, that means that your content may be used for future Gen AI model improvements. An example of what happens when you allow your content to be used for training a GenAI content model is that your audio might help refine models that create speech to text, which would help improve captions at Twitch but would also help improve captions across Amazon.”<\/em><\/p>\n

That seems clear enough, so what does that statement on X mean? My best guess is that Twitch is trying to claim that Amazon is using streamer content to train generative AI models, not Twitch itself\u2014a distinction completely without a difference given that Twitch is part of Amazon. I’ve reached out to Twitch to ask for clarification.<\/p>\n

The FAQ also notes that opting out of generative AI training “does not opt you out of all AI or machine learning uses at Twitch.” The platform can still use your data, including streams, VODs, and all the rest, for “AI-supported features” like auto-generated captions, although that is not retained to train models that can generate new content.<\/p>\n

This is hardly Twitch’s first big botch. If anything, the platform has been a hotbed of bad decisions over the years, on topics ranging from sexual content<\/a> to Donald Trump<\/a>, Palestine<\/a>, “politics and sensitive social issues<\/a>,” and when you can and can’t say “Zionist”<\/a>\u2014and that’s just within the past three years. <\/p>\n

But this move could be more impactful than most: Mike Futter, co-founder and analyst at game consultancy F-Squared<\/a>, said Twitch’s new generative AI training system has potentially far deeper consequences for streamers than might be immediately apparent: “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?”<\/p>\n

\n

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? This has deeper ramifications than it seems on the surface.<\/p>\n

\u2014 @futterish.fsquared.biz (@futterish.fsquared.biz.bsky.social<\/a>) 2026-08-12T22:17:07.463Z<\/a><\/p><\/blockquote>\n

“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. <\/p>\n

“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.”<\/p>\n

“We’re a staunchly anti-AI studio, and this now creates a real threat to our creative work,” added Futter, who’s also director of operations and publishing at indie developer Causeway Studios. And while he thinks Twitch and Amazon will face some pushback from game makers over its gen AI training plan, “this is David and Goliath. We’re going to need a critical mass of studios and creators to be loud about this.”<\/p>\n

Given Twitch’s history of switchbacks, I wouldn’t be surprised to see this new feature dropped because of the backlash\u2014but given Amazon’s determined embrace of generative AI<\/a>, I also would <\/em>be surprised by it. Either way, the bottom line at this point is that the new generative AI training model is a mess for Twitch on multiple levels, and it seems more likely to get worse than to go away anytime soon.<\/p>\n

\n
<\/div>\n<\/p><\/div>\n