Link to post: https://mastodon.social/@nicklockwood/117069670583842298
Short version: Two seperate “vibe-coders” used Claude to create an app–which turned out identical. And then the coders went after each other.
Transcript
Nick Lockwood: Holy shit, so the “Claude plagiarised an app” story is actually “Claude produced the same app twice when prompted to do so by two separate developers” 😅
Colin Cornaby
infosec.exchange/@david_chisnall/117066183348062085
I looked at the replies to this post, did some more digging, and found the actual story is even weirder.
BOTH projects were generated using Claude. So you have two people - both generating the same app from the same training data - having a public throw down over who copied who. Kind of embarrassing.
Want me to point out the commit I did to my app where I fixed the problem where it routes people to random fields in Mexico? That one took me several nights to fix. Yours also has it! Weird right
The bug that took him several nights to fix:
I typed into my computer “apt update && apt upgrade” and it improved a lot of the software on my computer.
If anyone else typed that into their computer, you’re copying me and I’m going to sue you!
bash: apt: command not found
Phew, lucky me. I always type it as two separate lines.
Still seems derivative 🤔
Ok, now that’s fookin funny. You love to see it.
It’s going to be so great as this tech moves into warfare, and atarts plagiarising war crimes from previous eras.
My theory is that a number of “devs” spent thousands upon thousands of dollars worth of tokens to walk Claude into building small segments of code. Then when these two vibe coders input very similar prompts, Claude used some of the previously generated code to serve up the same solution to both. Efficient, unless you want to profit from the slop code.
As time goes on, all Claude is going to do is throw previously generated code, others paid Anthropic for to “develop”, into its own blender, and serve it up as as some weird Frankenstein’s monster software.
If you are a developer, it might be time to start patenting ideas because copyright is fucked.
No, the other app was published on GitHub just a few weeks before, it can’t be inside the training data.
It’s more likely this: Claude code can and will do web searches to integrate the prompt if too generic or dubious, found the competing GitHub repo with the same name, got the claude.md with all the description, features and inner details, and rebuilt the copy on that.
Do we really want patents on open source projects that help track the quality of a night sky for photography?
Is there really an issue if such an application just is slopped out there?
I’ve been saying that for years: all LLM code output is “tainted” in terms of copyright. At the very least it’s all copyleft because copyleft code was used to train it, or more likely it’s just impossible to use legally at all because it’s conflictingly-licensed stuff smushed together.
From the viber’s blog: https://blog.terrygodier.com/2026/08/09/mea-culpa-dark-hours.html
A developer who created another web app called DarkHours.app replied to my comment on Bluesky yesterday to show me how similar the project was to his, including the name.
I think what’s so hilarious about the whole AI thing is how it seems nobody has hear of a template.
Like, the ads say “let claude make you a customized fitness plan!” when there are like a thousand tables you can download where you just fill in sets and reps or whatever.
The fact that we needed an AI to create two identical pieces of software is just such a waste of resources. Certainly there’s some open-source version of the app out there somewhere. Or at least enough to lazily glue together into the app.
Certainly there’s some open-source version of the app out there somewhere.
… which is what Claude was plagiarizing to build these ‘custom’ apps.
I had an intern this summer working on a project that involved mapping two data sets based on a description field. I showed him how to write a Python script using a fuzzy matching library to compare the two data sets and find matches. We then talked through the downsides of using a contextless string comparison and the situations that create false positives and false negatives.
We then tried to use a few of the AI tools to match them, hoping to demonstrate how a tool that understands context can do a better job at the expense of far more computation. All they would do was write and run a Python script using the same library, but every request would start with writing a new, slightly different script. Incredibly wasteful.
It turned into a great demo of how even if it returns a useful result, it’s rarely better than a person could do in a few minutes and is far less consistent.
Certainly there’s some open-source version of the app out there somewhere. Or at least enough to lazily glue together into the app.
Yeah, but that code is probably copyleft. They probably want the LLM to launder the license for them so they can get away with using it in their proprietary commercial product.
Well, sloppers were never smart to begin with.
Management types don’t understand how things work, so they make bizarre decisions
nobody has hear of a template
Eggsfuckingactly. Company says use AI as a starting template! Uh, hello, what if we used a fucking template as a template?? And that way we could have a deterministic source?? And we could track and fix bugs in the template?? No? We want to roll the dice each time to create the template? Ok. Ok…
I’m convinced that 90% of the people who think AI is amazing just never learned how to properly use office software. I saw someone insisting that it was amazing AI could send letters to hundreds officials all across the state to advocate for removing Flock cameras, saving them the time of doing each one individually. You have never heard of form letters? You think every letter you get from Chase Bank has someone typing it?? Did they just give up on teaching this shit to kids?
I saw someone insisting that it was amazing AI could send letters to hundreds officials all across the state to advocate for removing Flock cameras, saving them the time of doing each one individually. You have never heard of form letters?
That’s what they’re impressed by? Fuckin’ mail merge?!
That was being touted as the “killer app” for home computers in 1990!
I’m aware of doing a mail merge but don’t think I’ve ever done one. I’m assuming I could find instructions if I ever needed to do one. They might’ve even taught us about it in school; I remember in eighth grade a class where they taught us some computer basics using Microsoft Works. That was almost 30 years ago, though, and what I’ve seen with my own kid is very little computer instruction, and the schools all just give them Chromebooks and Google Workspace accounts. It’s all much more basic, so between that and using phones it seems like they actually have less computer knowledge than prior generations. Maybe Gen X, Millennials, and some part of Gen Z will have a peak percentage of their population with computer literacy, more than the generations before or after. I’ll be curious to see what happens as Gen Z and Alpha enter the workforce. I’m starting to take time to teach my kid more about using software more advanced than what they’re exposed to in school.
Your comment reminded me a of this song.
I must have missed the part where they taught the art of private equity
I’m thoroughly an adult (somehow?) but when I was at school, they mostly taught us the absolute basics over and over and anyone who hadn’t picked something up when they were five probably wouldn’t by the time they left school. It probably doesn’t help that the kind of people interested enough in computers to learn to use them properly and well enough to explain to children could also get a much more lucrative job doing something with computers than teaching children.
Yes but, if you do that you don’t get to completely give away your thinking faculties to the magic misinformation machine 🤡
I love how it keeps boiling down to that IBM quote about not letting computers make management decisions
