this post was submitted on 26 Jun 2026
148 points (92.0% liked)
Technology
85873 readers
3994 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 3 years ago
MODERATORS
Nooooo, you can't train on OUR data! That's illegal!!!1
The thief cries "catch the thief!"
Okay, so Anthropic distills MY copywriter data and it’s fine.
Alibaba distills Anthropic non-copywritable and that demands retaliation at the nation state level.
Fuck off. The rules are abundantly clear.
Anthropic just knows it has political leverage, that's it. Large parts of US infrastructure and business now rely on Claude models to even function the way they do with the staff they do, and you can leverage that dependence into political action and retaliation pretty easily in the name of "national security".
Lol corporate thieves bitching about other corporate thieves is the funniest part of 2026
New Qwen release incoming!
Q: if you steal a stolen thng, is it stealing?
you can't just call anything you don't like "an attack"
I declare an attack!
You cannot just declare it. That doesn't do anything...
Stop attacking me!
How about "terrorism" or "national security threat"?
Exactly!
Still, stop attacking me.
oh no, the data I stole is being stolen, whatever shall I do.
In other news, does anyone know a good source for crocodile tears? I ran out.
Sounds like Google complaining about scraping.
They stole to monetize without paying in money or attribution what we stole to monetize without paying money or attribution!
Well, they did pay, just after the fact.
Well they didn't pay me. But still used all my open source mit and agpl, gpl icensed code to train their model. And now I need to rent their compute back.
You chose to publish under that essentially says "do whatever, I dont care". I can understand people who wrote GPL code being peeved, but writing stuff under MIT is pretty much designed to let companies take it and not give back.
Sorry also some projects are agpl or gpl indeed. I actually meant to say agpl. They most likely also stole that. Let's be honest here.
For example my WineGUI project is agpl. Also I'm also fine if they trained on it. But then they should open source their models and data and their other projects.
I mean sure, Anthropic are pricks, but "they did exactly what the license I put on my code said they could" is probably not the way to highlight that.
My bad. I actually meant agpl. And gpl. You're right to point out the that mit is fine. Which is true. Mit can be used for whatever.
But my for example my WineGUI project is agpl. Pretty sure those Ai companies used all open source code regardless of the license.
I'm actually also fine if they trained on my agpl code. But then they should have open sourced their models and data. And their whole project.
That's settlement, not a payment for the stolen goods. If I steal your shit, and to avoid going to jail I offer you some money, I am not paying for what I stole, I'm paying to avoid further consequences.
What's a suitable punishment Anthropic?
(chasing) Whats a suitable punishment motherfucker?
Is there a scenario in which they both lose? I'll take that.
Alibabas Qwen were among the first open weights models that were actually useful and can be run on consumer hardware without too much difficulties.
If they continue with that, they will hurt the business model of the big AI companies significantly, accelerating the burst of the bubble.
Claude's still there, seems Alibaba's attack wasn't really all it's cracked up to be.
Now the US gov's attack seems to be working since Claude Fable 5 is still not there.
The article is not clear on what a "distillation attack" is... what exactly is Alibaba supposed to be getting away with here? The article mentions using many different connections through obfuscation networks and proxies... so that would get them around rate limiting, and maybe enable them to submit many queries on free accounts... just spin up a new account whenever you hit the token limit of an unpaid account. So basically it's a terms of service violation?
I don't see why it's necessarily a huge leg up for a competitor... they are just using the outputs of another model as training data. They still need to train their model, which is the expensive and energy intensive part.
It sounds to me like Anthropic just wants the US Government to help enforce its TOS internationally and force Alibaba to pay for those precious tokens? Because apart from that piece, the "attack" just seems like normal use of the service. If Anthropic's service has an inherent vulnerability, that's their problem.
Of course all the other comments about how they stole all their training data in the first place are spot on.
Distillation allows you to make a smaller model that can produce the same outputs as a larger model. Basically they're pirating all of the hard work anthropic did pirating the entire internet.
Alibaba gets a model that produces basically the same output for a tiny fraction of the cost to operate the model once it's finished training. Distillation training also uses basically all of its data from the big model (afaik it's all of it sourced from the parent model)
It's like if you took a lump of metal and showed it Porsche 911s until it turned into a 911 shaped chunk of metal that had 95% of the performance, but it only cost you $3000 for the ingot, and also cost ⅕ the amount in fuel and maintenance.
Ok, thanks for the detailed explanation. I guess if your goal is to make your model sound like another model that makes perfect sense.
I heard there was some new AI model that was so amazing for cyber security that they had to limit access to it. It's just too bad Anthropic couldn't use that.
Maybe they can protect themselves with a.. great firewall of USA?
Ai cloned the world. Please give me a break.
Enclosure or inclosure[a] is a term, used in English landownership, that refers to the appropriation of "waste"[b] or "common land"[c], enclosing it, and by doing so depriving commoners of their traditional rights of access and usage.
What's the science behind cloning?
LLMs are trained by taking a passage of text and masking out the next words. The LLM has to guess what the next word is going to be.
If you use the output of a fancy ass billion dollar model as your training data, you can duplicate the output style and "knowledge" of the parent model if you show it enough responses. That's basically what Alibaba did. They prompted the shit out of Claude and used the responses to train their own model which allows you to piggyback off of Claude's hard work pirating the entire internet. Your cloned model can also be smaller and leaner, being cheaper to operate.
I said this elsewhere but it's like taking a block of metal and showing it Porsche 911s until it turned into a Porsche 911 with 95% of the performance, and it also costs ⅕ the cost to maintain and fuel it.
here's the thing with the Porsche analogy. you had to buy or rent the Porsche first. paid for it and used it exactly within the TOS outlined in the contract. no law was broken.
what Anthropic is arguing is that anything their model comes up with remains Anthropic IP. this means they will literally need to sue every single one of their customers first, before they even have a snowballs chance in hell of pursuing Alibaba.
they already set the precedent by not legally pursuing their customers that use paid content generated by their model, and it automatically becomes the property of the end user.
95% of performance is impressive for a clone
It's approximate but yeah you can get roughly in that ballpark. The biggest benefit is making the model weights smaller and cheaper to run. You can fit 5X as many instances on the same server if you distill down while having basically the same output.
The main caveat is you need to absolutely hammer the main model with questions from all angles to try and get it to present as much of its internalized knowledge as possible. Which is why Anthropic is pissed about this since they're barely making money off of these prompts to train a more efficient competitor (BTW this is how "mini" or other models are trained. They're distillates)
AI going after their few paying costumers