This is super good for open source, right?

[hipsterelectron@circumstances.run]

@jzb @tante if it's a joke it seems rather like a rape joke to me

[jmj@hachyderm.io]

@tante @bagder but now you can just copy/reuse their clean room implementation because it not copyrightable.

[crazyeddie@mastodon.social]

@tante I was just reading not 2 weeks ago about a project that was completely rewritten and relicensed MIT or some shit. Like literally looking at the github repo.

So it's happened. Just not at scale yet.

[crazyeddie@mastodon.social]

@nihkeys @tante We've known the Poe effect for decades now.

[djgummikuh@mastodon.social]

@crazyeddie ? @tante

[8r3n7@mstdn.ca]

@tante Isn’t this just a fancy form of find-and-replace? You look for the license you don’t like, and you replace it with another one? You run the code through an obfuscator, which changes all the identifiers?

What I mean is, who says this isn’t happening already? OK, so this is an attempt to *normalize* it and accelerate. But if you can reverse engineer proprietary code, and use that as training data, or just do it the tedious way, who’s to stop you, and who can tell the difference?

I don’t know if these masters of the universe have considered all the legal and business implications of this wider project to destroy copyright, quite apart from the technicals.

I suppose they believe it’s a race. They are faster, thus guaranteed to win. No one else could possibly represent a competitive threat. What if they’re right? (One pictures hares and tortoises, but it’s hard to know when that fable is an appropriate metaphor.)

[ohno@ap.dny.social]

@tante @crazyeddie Yeap, this went down with chardet. Cat’s out of the bag. Now it’s just a matter for someone to “at scale” it.

There is a nice analysis of that debacle on https://writings.hongminhee.org/2026/03/legal-vs-legitimate/.

[hopeless@mas.to]

@tante I am not sure how much you have been paying attention, but the large tech companies gorge themselves on liberally licensed FOSS and very seldom does any money make its way back to the authors.

This has always been the case and no AI needed.

[nf3xn@mastodon.social]

@8r3n7 @tante Indeed using an AI would be an order of magnitude worse since you would also have to worry about the AI "helpfully" refactoring your code or just dropping large chunks of it, whereas your obfuscator is fairly deterministic.

[negativek@infosec.exchange]

@hopeless @tante Google has a public policy (with FUD) banning AGPL licensed code. Now they can just BS their way around that.