Submind YouTube summaries
Thumbnail for Twitch Updates Privacy Policy for AI Training

Twitch Updates Privacy Policy for AI Training

Watch on YouTube

Video summary

On August 14th, 2026, the streamer begins by addressing the extreme heatwave affecting his daily routine, explaining how he has been forced to shut down his powerful gaming PC during the day to avoid overheating his home office. Consequently, he has adjusted his streaming schedule to operate only after sunset when temperatures drop, a decision that temporarily halted his usual workflow and required him to rely on his phone for tasks he normally performs on a computer. This personal anecdote serves as an introduction to the broader topic of the day: Twitch's recent update to its privacy policy regarding artificial intelligence and data usage. The core issue discussed is a significant contradiction between Twitch's public statements and its updated legal documents. While the platform recently introduced a settings toggle allowing users to opt out of having their channel content used to train Amazon's generative AI models, the new privacy notice explicitly states that user data is still utilized to test and improve technology, including machine learning and AI. The streamer highlights comments from Twitch executives who admitted that an opt-in model would have resulted in zero participation, implying that data usage was inevitable regardless of user choice. This revelation suggests that even with the toggle enabled, users are not fully protected from their data being used for AI training purposes by either Twitch or Amazon. The video further explores the widespread nature of this practice across the tech industry, noting that major platforms like YouTube, TikTok, and X (formerly Twitter) similarly use user data to train their respective AI models without explicit, effective consent. The streamer argues that while features like auto-moderation, real-time sponsorship assistance, and viewer discovery already rely on AI, the controversy specifically centers on generative AI, which creates new content by learning from existing art and creative works. He concludes that relying on privacy toggles may provide a false sense of security, as data scraping and AI training are pervasive realities that affect almost every digital platform, urging viewers to assume their data is being used for these purposes unless proven otherwise.
Read the full video transcript
Good morning, good evening, good afternoon, everybody. Welcome to the stream. Thanks so much for coming by today. It's August 14th, 2026. I hope you're having a good day. If not, it's going to get better. Welcome to the stream. And thanks so much for hanging out with us. The crazy month of August. It's been hot. And that that's that's putting it lightly, chat. It's been hot. So, if you're wondering where I've been, like I have fixed my sleep schedule, right? And I I was waking up at the normal time, going to bed normal time. Problem was is that I could not stream during the day because I it was too hot. And I do have AC. Imagine that. It was too hot. It get it gets like 95 I estimate around 95 because the rest of the house was like 80. So, it was very hot in this office when I run my machine and I and when I stream and stuff. And you know the 4090, how that is, the Nvidia video card, the 4090. You know how that guy is. And it the becomes a heater. Those of you who use your PCs as a heater in the winter time know what I'm talking about. It becomes a heater. So, can't have a heater when it's already hot. So, I've been having to What my wife and I have been doing is that we've we shut down our PCs completely during the day. And basically, if I need to get online and do stuff, I have to use my phone. So, I I kind of put my work to a halt. So, I was like, oh gosh, cuz I don't I don't do stuff on my phone. I do everything on my computer. So, I decided I talked over my wife and I'm like, I might have to just flip my schedule again somehow because the only way I can actually work is if if it's not if the sun's gone down. So, that's what we're going to do. I'm going to be starting the stream around this time uh for the foreseeable future until it gets cooler outside. That makes sense. That makes sense, chat. Am I Am I talking sense? Am I crazy? I'm crazy, but am I Does that make sense? Because that's that's what's going on. That's why there's been a lack of streams, consistent streams lately. Mainly due to heat. Um yeah. So uh I have a new pin message thing and I also I made a widget today. Let me see if it's working. It's Oh, I got to I got to actually do the command. Um I made a new widget in Melt Spark today. It's supposed to automatically work when I pin a message. Let me unpin that. But I don't think it does. Let me try it again. No. But I do have a a manual override. So if I do pin and then a message, boop, it puts a pin message at the top of the screen. So yeah, have yap topics. Today's yap topic is about Twitch. Twitch has been hot in the news lately, dude. They have been hot in the news. They updated their privacy notice. And this is Zach's article or Zach's post about the article. He says they added information on how it uses personal information to test and improve technology, including machine learning and AI. And uh you probably been following the story since Twitch recently put out a new flag on their settings to where you can opt out of training Amazon's generative AI using your Twitch data. They gave you a a a toggle to turn that off so that you won't train Amazon's generative AI model. And uh Twitch did a patch note stream uh with the community manager and I think the COO or I forgot what that guy is. He's a product officer chief product officer. And they were talking about the issue and basically they were getting destroyed. The Twitch chat really didn't like what they were saying. Everybody's like, "We hate this." And a a a a notable thing he said, this was uh Mike Watson's face, a notable thing he said was he said, "We made it opt out because if we made it opt in, none of you would do it." And when he said that, I was like, "He actually said that." I audibly said it out loud. I was like, he actually said that. He actually said that. He said, if we made it opt-in, you would not do it. I was like, okay. All right. I don't think he cared at the moment. He was just like, you know. But anyway, this particular thing states the following. And Zach did a good job of highlighting it. He said that uh how Twitch uses personal information. Has a lot of information here, but what's highlighted is that uh and to test and improve technology, including machine learning and artificial intelligence, to support and enhance the customer experience across our products and services. This includes data to develop and deploy our generative AI models and services. So, the way this this thing is is worded is that Twitch is using your data, and these are the things they use your data for. To test and improve technology, including machine learning and artificial intelligence. So, this is a this is a direct contradiction, direct contradiction to what they said in that patch notes um in that patch notes uh thing. And what they said on on X as well. They said that Twitch does not use your data for generative AI. Amazon does. Well, this particular thing said Twitch says Twitch does. So, somebody lying, chat. Somebody's lying. Um and it's it's kind of sad that it's come to this, but I was under the assumption, and I'll show you what I posted about this. I was under the assumption that they were already learning and using our data. Right? So, if you look at this, this is what I said. This is when it was was first rumored. This was back on July 31st, where Zach says that Twitch and Amazon are apparently planning to use streams to train generative AI. And I said, "Between chat auto mod, which is the auto mod bot that sits in your chat, I knew that uses AI. That was like one of Twitch's first things that used AI was auto mod that I know of that was public facing, right? They probably have AI stuff in the back back end all over the place. But this was the first first public facing thing that I saw. And the new Twitch clip that feature, which is where you audibly say Twitch clip that and it just clips, you saw I saw that I was like, "Well, it's listening to everything you say like Siri and and and and and Alexa." So, that's using AI. I said, "Between those two things, it's safe to say that this is already happening, meaning that they're using your data to train generative AI. I'd love for all this to have an opt-out, but wouldn't surprise me if it doesn't. This was the before it was announced, right? So, I was under the suspicion that it was going to do that. And then, once we did get the toggle, they have this message on the settings page, right? Where it says, "Uh allow your channel content to train generative AI content models at Amazon. Turning this off does not opt you out of Twitch and Amazon using your channel content for other purposes described in the Twitch privacy notice, which was just recently updated, including using AI-supported Twitch features that benefit the community by facilitating streamer growth and monetization, such as real-time sponsorship campaign assistance." This is a new feature they offered. I don't know if you guys have heard of that, where if you they see that you're doing a sponsored stream and you're doing something wrong Clippy will pop up and be like, "Hey, you probably should put #ad in your message or you forgot this callout or why are you so stupid, streamer?" I don't know what they're going to say. But, real-time sponsorship assistance is something that they're doing with AI as well. Viewer discovery such as recommendations, which we kind of knew they were using AI for that in the back end. And community safety such as auto mod, which is what I mentioned in my post is auto mod is a bot that will automatically mod for you so you don't have to worry about having mods, right? Uh so, that came out I think that was that came out when they came out with the toggle. So, all in all it's it's it's one of those things where if you look at it from a pessimistic point of view if you look at it from a pessimistic point of view, you can safely say that your data on Twitch is being used to train AI. Whether you opt in or not. I think that's a safe assumption to make. Some people may challenge that. They may they may dance around that. And I'm not even saying if it's good or bad. I'm just saying like a factual thing that they are using your data. I think you need to assume that. If you don't assume that and you trust the toggle I got some bad news for you. They're using your data. As is most everyone else. There's a lot of uh people that are are challenging this and say, "Well, YouTube does it as well." Yeah. They do. YouTube does do it. Oh, well, TikTok does as well. Yes. They do. They absolutely do. Oh, x.com and Twitter, they use it. They use it for Grok to train. Yes. They do. They absolutely do. I remember when uh when Twitter announced that Grok was doing that, I immediately went in my settings and I I turned it off, but in the back of my head I'm going they're probably going to use it anyway. Uh and with YouTube they have a very similar thing as well that, you know, they want to enhance your audio and some other things that use AI and they said they need to gather your data to train so that they can better provide for these services. That that's that's using AI to use your data to train. So, yes, a lot of other companies are doing it. Welcome to now. A lot of us have been saying this for 5 plus years. Ever since this AI boom is like beginning to take off, like it's happening everywhere and people dismissed us when we were like saying, "Hey, they're they're using your data without your permission. Ding ding ding ding ding, we're ringing the bells. They're using your data without your permission. They're using your data without your permission." They're using your data without your permission and you need to do something about that and we got called all kinds of stuff, computer conspiracy theorists and nut jobs and all kinds of names. Oh, turns out it was true. Oh, okay. No, they say, "What's the difference between a conspiracy theory and the truth? About 6 months." Pizzama says, "I'm thinking that Twitch is now now being told what to do by Amazon and that Twitch has not much to say anymore." Um I do feel that shift as well. It feels like there's a shift in uh authority. Like somebody at Amazon said, "That's enough. I'm running the show now behind the scenes and you're going to do this." Or it just could be Clancy putting his foot down saying, "This is what we're doing." I'm sure they're using AI to to add close captions to clips. Yes, they are. Absolutely. Yep. Which is a good feature, by the way. It works pretty well. It's it's very close to what I see on some of these paid services out there that automatically caption your video using AI. It's very close to that. Even shutting it off in the Twitch setting, your data is still being used to train AI for other companies that are scraping public data on the internet. Yes. Yep. And yeah, there's a I forgot the name of the site that was basically taking people's VODs. They were they were I think you had to go to their site and opt out, but you could look up your name on this site and they would have your all your VODs up there. So, got to be careful. Yeah. And I use AI here. I just told you guys something I use it for. If you look at this, I've got a pin message up there. My my chat pin message is right there at the top. I just made that widget using Mel Spark, which I'm I'm I'm part of their content creator program. And they I'm promoting AI. So, I do use AI for for some things. I'm not going to sit here and say I'm anti-AI because I use AI for things. This widget right here, this weather widget, I created that using Mel Spark Pro. This coffee goal widget over here, I used that I created that using Mel Spark Pro. I put a prompt in and it just just spit it out. So, there's there's quite a few things that I have on my streams in terms of overlays and stuff that was created with AI. So, I do utilize it. My favorite my favorite way to use AI is search summary. Like if I go to Brave and I search for I don't know, let's say So, if I search for I don't know, uh recipes for bread pudding. This is my This is my favorite way to use AI. You get a AI search summary. Now, a lot of people complain about Google's AI search summary. On Brave is optional. You don't have to see this. Right? And I think Google's AI search summary is terrible compared to this one. So, what you see here is a summary of some bread pudding recipes. And you see there's a little link icon here? They'll give a summary or something, and you click this link and it gives you all the sources. And they have like some more here. And then you can just click elaborate and it takes you to Brave Ask, which is kind of like a little chatbot thing, and then it goes into detail. So, this is a this is this is my favorite use of AI because it summarizes search data for you for which you can go in and drill down. Because that that was how search actually started. Search results actually started that way where you had See how this has like a little summary of all the things here? They were actual snippets from the page. So, what they're using AI for is to take these snippets and consolidate them all, and then you can go through either look at this or you can go down here and go to these search results, right? So, that's my I love this for AI. This is something I I use every single day. So, AI I think AI definitely has some good uses, right? I think the the big controversial thing is that generative AI. But, yeah, that's that's that's the story with Zach and uh and Twitch. If you if you think that checking that checkbox is going to protect you from generative AI training or any AI training at all, then you're mistaken, I think. I think the big controversy around AI I was talking about this earlier. That's why I That was my segue. Showing off search. Generative AI takes art and takes creative things and it use it learns from it and it regurgitates it, and that that rubs people the wrong way. So, search results to me is not generative AI. Generative AI creates images and videos and stuff based on your prompts, and it trains on actual artists. And the controversy is that it trains on those artists without their permission.