My how fast the AI tech bros move to protect corporations but don’t move to protect the average coder, artists, writers.
“It’s so powerful! You’re only allowed to use it how we want you to”.
Allowed eh? How about go fuck yourself?
Hot take:
This is a good thing.
Claude Code is the “lowest common denominator.” It’s the default, zero research option. Its where you go if you follow popular trends.
If you actually want to reverse engineer, help code a game or something, there are hundreds of capable, uncensored LLM APIs to do that. Hacker communities will eventually figure that out, and finally realize Claude isn’t unique, and that Anthropic is toxic.
But for lazy spammers who don’t know that and aren’t motivated to seek stuff out… well, if this makes their life more difficult, good. If this sabotages Anthropic, that’s good too.
Don’t worry! Its STILL okay to reverse engineer the work of photographers artists writers actors voice over musicians modelers animators academics etc etc (feel free to add. the exploitation is literally everywhere!)
Only if you’re a corporation! If you’re a filthy worker then you have no rights.
Jokes on them when they face the consequences of usurping the jobs of all these professions, causing mass exit, and replacing with their error prone output of their slop and unoriginal thinking/artistry. I’m excited to see this outcome, not going to lie.
Another reason not to use a closed models
Misanthropic: Characterized by a hatred or mistrustful scorn for humankind.
Is this in response to that new “reverse engineer anything” tool from the people that reverse engineered the Adobe suite?
I’m really tired of the lie that ArtCraft is “reverse engineered Photoshop”.
It’s a slop factory. If you don’t believe me, download it and run it yourself. It’s just a frontend for tons of image and video GenAI bullshit. It doesn’t resemble Photoshop in the slightest, that narrative is pure marketing.
And it’s only “free” in the sense that you can download it for free. The first thing they ask you to do when you launch the program is buy tokens from them.
Even if it did work, Adobe’s legal team will obliterate them. I’m also not sure about the legality of using it for any paid work. It’s a moot point, though, since it’s slop, and better off using a cracked version that at least works.
of course. anyone who thinks AI can be used as a tool for independence is deluding themselves
anyone who thinks AI can be used by us commoners for much longer is deluding themselves. it is so obvious that these big LLM companies want to capture the market by making everyone dependent and then crank prices way up.
The adobe suite was not reverse engineered, in the sense in which the term is applied to binary, closed-source softeware. Rather, it was cloned.
This is technically and often also legally very different. Cloning is a new implementation of a system which reproduces the abstract API of an existing System. Since APIs are not subject to copyright law, this is legally unproblemaric. It is also a creative process (so far). For exsmple, Linux is a clone of AT & T Unix.
Reverse engineering uses techniques like analysing the internal behaviour and decompiling the binary code of an existing product. It is im general not illegal, but it can violate license agreements and is more in a grey area. Laws like the DMCA might also apply if some kind of copy protection is used.
BTW, many Linux drivers are reverse enginnered to some degree.
For exsmple, Linux is a clone of AT & T Unix.
Unless your definition of “clone” is very different from the mainstream one, you’re talking out of your ass here. Linux was written from scratch. The only thing they’ve got in common is being POSIX-compliant.
Yeah, it clones the Unix / POSIX API: Programs written for Unix will run on Linux. In the same way, the Photoshop/Office lookalikes are clones, they were written from scratch to yield a result compatible to the original.
Look at the explanations I linked.
I’m hopeful this is positive, they’ll start loosing a lot more to Chinese models and I can assure you local models will blow up too.
I can assure you local models will blow up too.
Why?
They are really good already, even if they have some kind of censored limitation, once you jailbreak it js jailed break for ever The only limitation right now is the poribitive gpu prices
To those who say “but what about game preservation?”:
GenAI slop ruined communities around game preservation. And for every time a AAA dev getting this treatment, there’s at least 3 indies getting it worse.
Yep a lot of decomps just get a slop coded version and people stop caring about it because the slop version is “good enough.” Most people doing the decomps it’s also a hobby project that they’re doing for fun as well, the end goal is just a bonus.
There’s already a meme that a lot of those decomps are just emulators + ROMs in a single repository.
So this is how these companies are going to safeguard against IP theft lawsuits! Leave our theft be, or we’ll tell everyone how to get in.
Also it’s incredibly telling how Anthropic doesn’t add code to stop suicide ideation in teens (what they are their ilk are being sued for) but can somehow very quickly stop reverse engineering. A tool that values the captial holders through & through. Their “democratization of intelligence” schtick is obvious now.
OpenSource frontier models ASAP! Obviously never gonna happen, but a fool can dream atleast.
Also it’s incredibly telling how Anthropic doesn’t add code to stop suicide ideation in teens (what they are their ilk are being sued for) but can somehow very quickly stop reverse engineering.
This.
It’s not ok when the commoners get some power back huh. What a shame, I’ve been loving playing my favorite console fps games as a kid, now with keyboard and mouse support.
When it hurts jobs, AI is the future, when it hurts rich friends needs moderation.
This was literally destroying indie games. Look at the deluge of slop that came out when ‘sort items onto shelves’ kicked off as a game type.
It’s interesting to see modern day “shovelware” is still around. The problem of shitty clones and rip offs of popular games has always been around. Or the classic bait and switch of advertising vs actual gameplay.
The proliferation of these things made possible through the internet and sites like itch.io and sped up by lazy AI slop has certainly made it even more pronounced. The difficulty of finding good indie games amongst a huge pile of trash has only grown.
That’s what late stage capitalism mixed with desperation, delusion, and the desire to get rich quick will get you though.
I can’t understand why anyone would play such a “game”. You’ll get more satisfaction and results from actually tidying your room!
Indies have been copied since Unity introduced the asset store, don’t fool yourself, this is made to stop the reverse engineering of big company games and apps, not the little guy.
Now only the people that have enough money to pay for the privilege get to use it!!! Hurray?
That was always the plan
Man, the one thing I was interested in, the decompiling of abandoned software to get old games and software back, is being limited.
The task Large Language Models are best at is pattern correlation. Out of all the possible applications, reverse-engineering seems the most reasonable use case to me.
But figuring out how stuff works is the exact opposite of what they want vibe coders to do: Don’t think, don’t train your own skills, just trust the machine. And if it doesn’t work, just prompt again and keep wasting more tokens, pushing the usage metrics to convince investors they’re constantly growing, burning the planet and your company’s money.
The “There’s no way we can stop our agents from going rogue” people, suddenly have a burst of ability to reign in their creation when it suits their interests.
Funny that…
Does anyone think that perhaps what they are afraid is that source code of such clone would be compared to the original and show evidence that maybe the model was already trained on the code of the game?
I think this is definitely a concern, yes.
I think it’s very likely that these models are effectively being tricked into retrieving source code of projects from memory of seeing the actual source code in the biases. This is why some project are much easier to uncover than others.
They aren’t being tricked. They are doing exactly what they were designed to do
Fair fair. Ih suppose what I mean is that it’s not really reverse engineering so much as it is recollecting from training samples.
I’m no expert, but if my understanding of LLMs as ‘probability calculators’ is correct, then if you prompted one properly to produce code which does x (a Photoshop function for example) and it had only been trained on one example of this (the Photoshop source code) it would reproduce the PS source code because that has the highest probability of that being the correct ‘answer’.
This is baasically correct, though what the models actually produce is a little more randomized than that. Though, if there aren’t any words that are anywhere cllose to being likely to follow next, then there won’t be a any other valid options to choose.
Where would they get the source code from so they can include it in the training data in this scenario?
Looks like everywhere companies mandate employees to install copilot or equivalent.
I think originally they were saying they won’t be using the proprietary code for training, but I don’t trust those companies when they already broke laws.
From the stolen math research for tha NS, from the company response that they can’t guarantee that the model didn’t train on the mathematician sessions, we know for sure they are using the inputs for training lol
This is full conspiracy from me, but could the terms of service for github grant the right to “enhance tools” based on private github projects. These then get moved around as ‘anonymized’ data to train off?
The models don’t store access to their training data. So no, there’s no source code in the model to retrieve.
It’s long been proven that LLMs can produce exact copies of their training data. Not every piece of training data, of course, but enough where spitting out exact replicas of proprietary code is a legitimate concern. Not to mention that copies don’t need to be exact in order to be a problem.
The models are essentially a giant lossy compression applied to the training data.
They don’t store the loseless version of the training data sure, but that doesn’t mean it can’t reproduce something close enough to it, it’s literally how the whole thing works.
My understanding of it is that it’s far more complex than “lossy compression”. Of course it can generate text that is close to something in its training data, but I don’t think people have the right mental model if they think the model itself somehow has whole chunks of code in it. It does not.
My understanding of it is that it’s far more complex than “lossy compression”. Of course it can generate text that is close to something in its training data, but I don’t think people have the right mental model if they think the model itself somehow has whole chunks of code in it. It does not.
So why can they recreate entire articles from the New Yorker and complete functions from the training set? Don’t fall for the slop laundering.
I’ve never seen any proof of this? They can fetch data and present it as their own maybe? But models don’t store training data verbatim. It’s like 6+6 and 4+8 both = 12, but they just store 12 and how it got there is unknown.
I’ve never seen any proof of this? They can fetch data and present it as their own maybe? But models don’t store training data verbatim. It’s like 6+6 and 4+8 both = 12, but they just store 12 and how it got there is unknown.
More power and value to the open weight models, cool