Do NOT add AI to Scratch — #NoAIOnScratch

EDawg2011

AndPherbCodes wrote:
#3062
AndPherbCodes wrote:
Also, I don't want to be that guy, but perhaps it's not for them to directly train models, but to make it so that if openAI yoinks our projects, we can't do anything to stop them, and in turn scratch doesn't have to protect us. It's bad, but a different type of bad.
plitd

What does plitd mean?

Last edited by EDawg2011 (March 1, 2026 20:30:29)

scratchcode1_2_3

ajskateboarder wrote:
(#3060)
StarFutureFox wrote:
UHH, if you're gonna leap off the demo vid and claim it doesn't show anything except an ‘AI assistant’, I'd like to remind you of something someone else (I can't remember who) said earlier: it created a random ah food truck that was not in any way related to the original project.
The demo video is all we currently know about the CLA and it was what scratchcode1_2_3 was talking about. The ToS contains information about what Scratch is allowed to use user Content for, not the functionality of their planned assistant. Scratch may very well create an assistant that doesn't implement everything they grant themselves in the ToS.

I'm not sure what food truck that person was talking about because I did not see assets being “generated” anywhere in the video. Please have a source or a video timestamp
scratchcode1_2_3 wrote:
read the ToS twin
I have already many times
also it can just straight up steal sprites………
Don't see how this is an issue

you probably just don't care about having ownerships of things and already live under the “Own nothing and be happy” rule that we are approaching.
Who cares if they don't implement what the ToS allows them to? They gave themselves the power to do so, they're probably gonna do so. Why would they explicitly give themselves such power if they won't use it? It only benefits them.

Foxofpeace

EDawg2011 wrote:
AndPherbCodes wrote:
#3062
AndPherbCodes wrote:
Also, I don't want to be that guy, but perhaps it's not for them to directly train models, but to make it so that if openAI yoinks our projects, we can't do anything to stop them, and in turn scratch doesn't have to protect us. It's bad, but a different type of bad.
plitd
What does plitd mean?

post got left in the dust. i created the acronym

StarFutureFox

scratchcode1_2_3 wrote:
ajskateboarder wrote:
(#3060)
StarFutureFox wrote:
UHH, if you're gonna leap off the demo vid and claim it doesn't show anything except an ‘AI assistant’, I'd like to remind you of something someone else (I can't remember who) said earlier: it created a random ah food truck that was not in any way related to the original project.
The demo video is all we currently know about the CLA and it was what scratchcode1_2_3 was talking about. The ToS contains information about what Scratch is allowed to use user Content for, not the functionality of their planned assistant. Scratch may very well create an assistant that doesn't implement everything they grant themselves in the ToS.

I'm not sure what food truck that person was talking about because I did not see assets being “generated” anywhere in the video. Please have a source or a video timestamp
scratchcode1_2_3 wrote:
read the ToS twin
I have already many times
also it can just straight up steal sprites………
Don't see how this is an issue
you probably just don't care about having ownerships of things and already live under the “Own nothing and be happy” rule that we are approaching.
Who cares if they don't implement what the ToS allows them to? They gave themselves the power to do so, they're probably gonna do so. Why would they explicitly give themselves such power if they won't use it? It only benefits them.

EEXACTLYYYY!!! That's exactly what I've been trying to say every time someone brings that up! I mean, imagine: I go to the store and buy a fifty dollar cake, right? The baker agrees they will give me the cake if I give them fifty dollars. Or, the baker gives me the cake for free. Y'know what, let's go with the free cake instead because you pro-AI people are gonna go ‘tHaTs a wRonG anAlOgy bEcUAse tHe AI trAiNIng dOesnT cOsT anYthINgggGG!’ So anyways! The baker is a very kind person and gives me the cake for free! I take the cake home and proceed to let it rot in the corner without even eating or trying it, because it makes perfect sense to intentionally get something and never use it!

And alsooo! Agreed! Why would you ‘not see an issue’ with AI slobbering over people's stuff and learning from it to generate slop? I mean, you're a pro-AI, so I guess you don't care, but other people care! It should register in your brain that some people do not like their stuff being fed to an AI ;-;

ajskateboarder

scratchcode1_2_3 wrote:
you probably just don't care about having ownerships of things and already live under the “Own nothing and be happy” rule that we are approaching.

I absolutely care about having ownership of things. That's why I took issue with third parties being able to train off user Content

With that being said, isn't your position that you would've wanted an assistant in some other form? Doesn't most AI training involve most of the same ethical issues?

Who cares if they don't implement what the ToS allows them to? They gave themselves the power to do so, they're probably gonna do so. Why would they explicitly give themselves such power if they won't use it? It only benefits them.

I didn't say they would absolutely avoid some things they invented in their ToS, just that's a possibility since their ToS contains tons of other contradictions. Do you think Scratch is planning to close source their website and editor, and remove themselves from Google because they ban “search agents?” Or that Scratch will ban users from countries “designated by the U.S. Government as ‘terrorist supporting’?”

StarFutureFox wrote:
That's exactly what I've been trying to say every time someone brings that up! I mean, imagine: I go to the store and buy a fifty dollar cake, right? The baker agrees they will give me the cake if I give them fifty dollars. Or, the baker gives me the cake for free. Y'know what, let's go with the free cake instead because you pro-AI people are gonna go ‘tHaTs a wRonG anAlOgy bEcUAse tHe AI trAiNIng dOesnT cOsT anYthINgggGG!’

I don't really understand this analogy but I think you should just read my above post

Exascerbescence

I probably shouldn’t quote myself like this but this is a perspective I don’t think many other people will be able to give and it’s important to me people see this

Exascerbescence wrote:
pom_cookie wrote:
cambegonefr wrote:
Bitebite12 wrote:
According to scratch, the AI will assist. The scratcher will still make the project, but the ai will help them fix bugs, brainstorm, and break down things you don't understand.
And that's the problem the OP stated. What about trial and error, creativity, and the good feeling of learning on your own?
This is a very good point! So many people lack proper cognitive function because of over using AI and not actually using their brains to learn and problem solve. Does Scratch really want that for all these kids on their website?
I did once get addicted to AI when I was a much more vulnerable person and I think the people who support AI should know what that’s like before they make their decision.

You lose all of your energy, you no longer have any will to create or do things, you stop researching and pursuing inspiration and ideas and thinking deeply and trying to understand the world around you, because why not just ask the AI? You lose the ability to fact-check and become gullible and eventually you stop having the energy to just go for a walk. This is not exaggeration, this is personal experience and I only got out of it because I knew the AI (or rather, the company behind it) was there to take all of my time and control me.

How in the world are young children supposed to navigate this? I do realize that Scratch is a non-profit, but you still quite directly risk exposing young children to these other intentionally controlling AIs. I have seen grown adults who allow themselves to get controlled by this sort of hyperconvenience and not even take a moment to question it. There are 7 year olds here. I fear that including an AI here would not just be annoying and saddening but could quietly ruin lives.

AndPherbCodes

Foxofpeace wrote:
EDawg2011 wrote:
AndPherbCodes wrote:
#3062
AndPherbCodes wrote:
Also, I don't want to be that guy, but perhaps it's not for them to directly train models, but to make it so that if openAI yoinks our projects, we can't do anything to stop them, and in turn scratch doesn't have to protect us. It's bad, but a different type of bad.
plitd
What does plitd mean?
post got left in the dust. i created the acronym

is anyone actually going to scream at me about how I'm stupid or something?

EDawg2011

Exascerbescence wrote:
#3071
I did once get addicted to AI when I was a much more vulnerable person and I think the people who support AI should know what that’s like before they make their decision.

You lose all of your energy, you no longer have any will to create or do things, you stop researching and pursuing inspiration and ideas and thinking deeply and trying to understand the world around you, because why not just ask the AI? You lose the ability to fact-check and become gullible and eventually you stop having the energy to just go for a walk. This is not exaggeration, this is personal experience and I only got out of it because I knew the AI (or rather, the company behind it) was there to take all of my time and control me.

How in the world are young children supposed to navigate this? I do realize that Scratch is a non-profit, but you still quite directly risk exposing young children to these other intentionally controlling AIs. I have seen grown adults who allow themselves to get controlled by this sort of hyperconvenience and not even take a moment to question it. There are 7 year olds here. I fear that including an AI here would not just be annoying and saddening but could quietly ruin lives.

@Sasha_Banks272 PLEASE add this argument to your original post! It'd STRONGLY support our viewpoint.

Last edited by EDawg2011 (March 1, 2026 20:31:16)

scratchcode1_2_3

ajskateboarder wrote:
(#3070)
With that being said, isn't your position that you would've wanted an assistant in some other form? Doesn't most AI training involve most of the same ethical issues?

Yes, I don't mind ethical issues. But this is Scratch we're talking about. Think about Scratch like your best friend. I don't mind big corpos because, well, they're a thing that exists and we can't really do anything about it. At least we know they're giving us service in exchange for shady data collection. But seeing Scratch do this is like having a sibling or a best friend you've had your whole life backstab you and throw you in a dumpster, then take your wallet and sell your house's furniture for their own greed. Then leave me stuck with a half-baked terribly-researched uninformed ignorant hallucinating excuse of an assistant and say “hope you'll be happy with this lobotomized clanker that I made using the money the evil billionares gave me”. It's much worse because it's a betrayal. It's not like Google which was a corporation from the start. They aren't trying to hide the fact that they're a business and that's what makes us somewhat respect these companies. Scratch is trying to hide under some non-profit facade when what they're doing is very corporate style. They're trying to downplay and make us feel better about something terrible they did that's literally so in our noses.

Instead of adding new blocks, improving the homepage, fixing cloud variables, so many things they could've done, they decided to take their so-precious money and shove it into a clanker as a distraction from the corporate move they pulled. The AI is a distraction. And it sucks anyways. The CLA sucks in so many ways. It's clear Scratch is barely informed about AI from these demos. I didn't want to say this in Scratch, but AI has caused deaths, hospital visits, serious money loss, and more, when it is implemented wrong. This is not a correct implementation. You can't just drop an LLM into a chat sidebar and call it an AI assistant. There's so many things you can do with an LLM, but if you don't know how it works, you're going to create a garbage-ridden slopbot and hope people like it, which, they won't.

Edit: Maybe this is what the Scratch Foundation wanted. By making a terrible AI assistant, that was what would get all the hate instead of the other egregious and unethical things in the ToS. You can say all you want about it being a conspiracy, but it ain't a conspiracy if it's in the ToS. If it's possible under the ToS, we must assume they will take advantage of it in every way they can think of. We can't trust them to keep their word under a legally-binding document that allows them to not keep their word. Think about it this way. Your best friend gets a shiny jack-of-all-trades warrant signed by the president himself that allows him to take ownership of anything he wants and sell it (the third parties) or take it apart for spare parts (AI training) and now he carries it ready-to-use everywhere he goes. Since he's your “friend”, he also gives you access to this cool gadget he frankensteined into existence using everyone else's liquidated assets (the CLA assistant). Said assets can be anything from kids' toys and drawings (beginner projects), to people's cars and inventions (Glue3D, raycasters, algorithms, multiplayer, risc-v emulators with working linux kernel or DOS). You can trust him that he's not going to use it on you and steal all your possessions, but you have to stay on high alert and hope for the worst, since he has a ridiculous amount of power that he didn't exactly need to keep living. But also, you can't just stop being his friend. Obviously you can, but it's very hard to do since you've been best friends your whole lives. So you either keep trusting him, or you move to a place that he doesn't have said power in (another website), but rather, another person will (literally every other corporate platform). The other person will feel better since you know from the start that he's evil, while your friend wasn't up until then. It's less exhausting because this other person isn't pretending to be friends with you.

Last edited by scratchcode1_2_3 (March 1, 2026 03:22:09)

-BluEagle-

I’m gonna be honest: I’m going to be heavily upset if scratch 4.0 is just this AI assistant and nothing else.
Children nowadays are more interested in technology than they were at the time that scratch 3.0 released. Giving them more things to tinker with like bigger, wider project players, usb gamepad code, etc would be really awesome. But if it’s literally just gonna be an ai that btw, could probably get a lot of frustrated 8 year olds very very quickly because not every single child knows how ai works, 4.0 is just not gonna be that cool and will probably be a bit underwhelming if anything.

Last edited by -BluEagle- (March 1, 2026 03:56:22)

ajskateboarder

scratchcode1_2_3 wrote:
ajskateboarder wrote:
(#3070)
With that being said, isn't your position that you would've wanted an assistant in some other form? Doesn't most AI training involve most of the same ethical issues?
…

Can't say I agree with everything in here but I really appreciate this post. I also do think there's other egregious ToS terms worth giving attention to

MiaELamanna

-BluEagle- wrote:
I’m gonna be honest: I’m going to be heavily upset if scratch 4.0 is just this AI assistant and nothing else.
Children nowadays are more interested in technology than they were at the time that scratch 3.0 released. Giving them more things to tinker with like bigger, wider project players, usb gamepad code, etc would be really awesome. But if it’s literally just gonna be an ai that btw, could probably get a lot of frustrated 8 year olds very very quickly because not every single child knows how ai works, 4.0 is just not gonna be that cool and will probably be a bit underwhelming if anything.

Frrrr

MiaELamanna

scratchcode1_2_3 wrote:
ajskateboarder wrote:
(#3070)
With that being said, isn't your position that you would've wanted an assistant in some other form? Doesn't most AI training involve most of the same ethical issues?
Snip

I very do agree with that,tbh the SF kinda betrayed us in some sort of way

pippy2011eight

MiaELamanna wrote:
scratchcode1_2_3 wrote:
ajskateboarder wrote:
(#3070)
With that being said, isn't your position that you would've wanted an assistant in some other form? Doesn't most AI training involve most of the same ethical issues?
Snip
I very do agree with that,tbh the SF kinda betrayed us in some sort of way

yes, they did with the “training our projects on AI”
And I would just like to say,
it was the SCRATCH FOUNDATION
not the scratch team,
2 different groups who work with each other.

Exascerbescence

I have a more technical argument against 4.0

2.0 happened because Squeak was very, very old
3.0 happened because Flash was dying and horribly insecure
React isn’t even close to dead

Even if there is actually the sort of technical change that the version naming would imply, the 2.0 pages that should’ve gotten replaced a decade ago are still around and still using a different framework to the 3.0 pages, which would mean now we would have three types of pages which need to be maintained by different groups of people, and as stated before React isn’t dying yet which means there isn’t really a reason to change frameworks.

So what I conclude is that this label (the jump from 3.0 to 4.0) is being abused to raise hype. This is also further supported by its known features (AI is obviously heavily exploited for hype and is the main feature here and membership is something that would likely excite investors), and it is concerning that the SF is trying to raise hype with the sort of things profit-motivated investors love

Last edited by Exascerbescence (March 1, 2026 05:13:37)

MiaELamanna

pippy2011eight wrote:
MiaELamanna wrote:
scratchcode1_2_3 wrote:
ajskateboarder wrote:
(#3070)
With that being said, isn't your position that you would've wanted an assistant in some other form? Doesn't most AI training involve most of the same ethical issues?
Snip
I very do agree with that,tbh the SF kinda betrayed us in some sort of way
yes, they did with the “training our projects on AI”
And I would just like to say,
it was the SCRATCH FOUNDATION
not the scratch team,
2 different groups who work with each other.

Ye the ST has only a little power and the SF has control of almost everything
Edit: YESSSSSS QUEEN OF THE PAGE

Last edited by MiaELamanna (March 1, 2026 05:10:51)

thunderzee17

AI can do good and bad things. For example: AI image and videogenerators can generate images and videos easily, But might Cause misinformation. They also don't care of copyrights of painters , such as people who is still alive.

StarFutureFox

scratchcode1_2_3 wrote:
~

I really love what you've said. I read the whole thing. This is all really important, I hope everyone who sees this reads every word.

Before I continue:

Instead of adding new blocks, improving the homepage, fixing cloud variables, so many things they could've done, they decided to take their so-precious money and shove it into a clanker as a distraction from the corporate move they pulled.

I agree! Sometimes Scratch even gets really really slow and it's hard to use, and my inbox takes way to long to load, and I can't type in the costume editor. And what about the always-around-problem of comment threads taking forever to load on profiles? There are literally so many bugs they could've fixed instead of doing….. this.

But this is Scratch we're talking about. Think about Scratch like your best friend.

This sentence… people need to at least read this sentence. This is Scratch, the goofy coding site that has now also become a place for art, animations, music, and more. It's used in schools all over the world. People use it all over the world. It's for all ages and it's safe and nice and it doesn't have advertisements and… what are we doing? It's non-profit! It's like a gift from god. I mean. What you mean by telling us to treat Scratch like a sibling or friend is- Treat Scratch like it's something you care about. Think of it as a good friend for a second. I know many of you, pro-AI people especially, have been thinking of Scratch as just a random website that you have an account on, or whatever, but just for a second pretend that you're a neutral person who has done a lot of things on this website. Imagine you have friends on this website. Imagine for a second that you love this website and that you really care about your work. Like, you really care about it. Got it? That's easy, isn't it! It's easy to imagine. Most people already have that ~ if you do, you're already half done. If you don't just try that. It's not hard, it really isn't - or if it is, you've got issues. Now, if you're the person who doesn't already have that, think about this: Now. Please think about AI, alright? I'll do this by facts, because most of the pro-AI people talking here completely avoid considering that emotions exist in their posts. So, first fact: AI is not a human. Second fact: AI has no feelings. Third fact:

AI has caused deaths.

AI does not give a rice grain about the things you've created, and it greedily feeds off of every piece of your hard work that it can find to create the same thing in seconds, except without the love. And some people (the number grows every day as AI gets ‘better’ and ‘better’) can't even tell the difference. And now Scratch, the one ‘social media’ that has steered clear of AI, is going to start using it. They've agreed to let AI models train off of every single project on this website. And, on top of that! Have you ever felt the great feeling of coming up with an idea, or fixing a bug, or doing even a small bit of something (code/art)? If you have, congratulations! You're a human. Now, the malevolent artificial ‘intelligence’ I described a few sentences ago is being integrated into Scratch, which I also described a few sentences ago. Now think how you would feel if you cared about Scratch the way I described you would, and if you cared about your hard work the way I described you would. Think about it, just think about it…

-BluEagle-

I think the CLA can have genuinely better uses
Like a translator
A guide for when scratch games become a bit too hard
(This is a suggestion I made just now) being able to describe code that can be hijacked that has been hidden (as a feature) in a way that makes it 1: not able to hijack the code and 2: still able to make the code visible in some form

thunderzee17

thunderzee17 wrote:
AI can do good and bad things. For example: AI image and videogenerators can generate images and videos easily, But might Cause misinformation. They also don't care of copyrights of painters , such as people who is still alive.

In scratch,it might be good for beginners who don't know how to code but might replace people's minds, not creativity . The AI also gets code from other people using scratch, like the painters.

Discuss Scratch