Hacker News new | ask | show | jobs
by goobatrooba 5 hours ago
One other Twitter comments reveals that they probably just asked an AI to copy Papermark. Evidenced by AI comments saying the page was aligned to the "reference"

https://xcancel.com/ffumarola/status/2070479755892371713#m

5 comments

Unreal. I had to go back to the original Tweet to confirm that screenshot wasn’t faked.

The comment clearly says “Mirror’s the reference design’s”

I don’t know how they could try to spin that as anything other than having an LLM launder someone’s code as a “reference design”

Even if they try to argue that the “reference design” was Figma, the identical copy means they had to have copied Papermark into the reference design.

The fact that they’re refusing to back down and admit they made a mistake is not a good sign for the company. I would not want to be one of their clients when it came to trying to dispute something.

> The fact that they’re refusing to back down and admit they made a mistake is not a good sign for the company.

I wish this were true, but the current political and corporate climate is that nearly anything is justifiable as long as you win, where winning is money or power. Fraud, corruption, extortion, etc.

> I would not want to be one of their clients when it came to trying to dispute something.

I find most b2b transactions are hostile, and the purpose of sales or customer success is to smooth over the hostility. Tremendously more true in the B2C space, and only accelerated by the aforementioned political and corporate climate.

In other words, as long as their staff is charismatic / crafty enough, this “scandal” will slough right off.

I meant it's not a good sign for anyone considering working with the company.

If they refuse to back down and won't admit errors on something this obvious, I would not want to be dealing with them on an insurance claim.

Problem is law hasn't caught up with how easy this previously GOOJF practice became with LLMs:

https://en.wikipedia.org/wiki/Clean-room_design

You might be able to argue that using an AI trained on open source or source available code is not a clean room implementation. IANAL.
Funny enough, I just published this blog post yesterday:

The Discoverable Evidence of AI-Assisted Software Porting

https://williamcotton.com/articles/the-discoverable-evidence...

I wonder if this will get worse with so many firms now using agents.

You’re sharing your entire code base with a 3rd party you have to trust not to train on it.

If they do your competitor’s just to ask it to produce something using your business as a reference.

Good luck taking that one to court considering what happened to the publishing industry.

Is that real? Imagine they've taken the code out if so, difficult to verify.
The era of chinese level competition is here.
If it were, they would make it worse then in time make it better, not just copy verbatim