Compiled software is no longer a safe hiding place for source code.
Using AI agents and only the shipped binaries and public documentation,
we reconstructed maintainable source code for Slack (client and server),
Apple's Finder, and the TinyTurla-NG backdoor, then added a feature Slack
does not have. We look at how far each reconstruction got and what
vendors can do to protect their software.
Plaintiffs can claim a violation occurred, and the fun will begin when discovery compels the defendants - the AI companies - to disclose what their models were trained on.
As to “there’s no real author to pin it on”, that doesn’t hold water legally: I can’t put a brick on my car’s accelerator, let the car slam into a building, then claim I wasn’t responsible for what the car did. Similarly, Waymo is held responsible for what their autonomous vehicles do on public roads.
Even having been trained on copyrighted data won’t necessarily prove that something produced by AI actually used copyrighted code. That might get the AI companies in deep shit, but I’m not sure it’ll cause blanket problems for all AI generated code.
As for responsibility, that’s a very different thing from copyright ownership. I understand the idea as I proposed it isn’t intuitive, but this isn’t a fair comparison.
Well, the whole subject is hairy as fuck - or if you want to put it kindly, everybody is in uncharted legal territory.
My original comment simply stated that claiming “everything becomes open-source” simply isn’t true. And whatever the courts decide ultimately, I’m pretty sure the AI companies will be sued by the Adobes and the SolidWorks whose ultra-expensive software products are suddenly obsoleted by free (as in beer) slop they don’t control and they don’t make a dime from, that’s just good enough for people to give the expensive originals the middle finger.
At the moment, the PhotoShop clone and the SolidWorks clone aren’t good enough to be a menace to the real McCoys, but they’re not far off. And at the rate they’re evolving, in 6 months, they’ll be better than the originals.
If you think Adobe and SolidWorks are going to take that sitting down, you’re dreaming: they’ll sue the AI companies with all they’ve got, even if they stand to lose, because they simply won’t have any other option to stay alive. And if enough of those software vendors under threat join in, they’ll sue the AI companies to oblivion.
And that’s not to mention all the vendors whose locks will be broken by AI, and who will sue the AI companies for aiding and abetting DMCA violations.
There’s 1,000% a huge legal battle brewing on the horizon. If it comes too soon, I think it’ll get shot down for national security or some other stupid bullshit to enable AI, but if the bubble bursts, they may be able to win and bury what’s left of the industry. Of course, then open weights will just take over because one does not simply close Pandora’s box.
But as to the point I was making and why I joined in, going back to your first comment in this chain,
My point is that I don’t think this is necessarily clear cut. There is certainly some unresolved ambiguity to copyright with regards to vibe coded content, mostly from my understanding based on the idea that purely AI-generated content doesn’t actually have a human author to ascribe authorship and therefore ownership to, but I really don’t think there’s a clear cut way to prove these clones are built on copyrighted code or anything else that would help you directly invalidate their licenses. They may have used a tool that violated copyright in its construction, but I think that’s a very different thing from being able to prove that specifically the code they published was somehow problematic.