Opinion
Theft by Proxy — When AI Becomes a Tool to Launder IP Theft
The story of a product I spent five years building, cloned with AI and released for free without attribution — the concept of "theft by proxy" and what we can do about it.
I spent about five years building a product. Not five years of code alone, but of understanding users, fixing mistakes, and redesigning again and again until it became something thousands of students relied on daily — and more than a thousand paid for.
Then, within days, a copy of it appeared.
What happened
A competitor used AI to "clone" the product's functionality, released that copy for free on GitHub, and promoted it to tens of thousands of students — with no attribution to the original.
It wasn't inspiration, nor fair competition over an idea. It was a transfer of the work's essence, except AI stood in the middle as a proxy, so the output looked "new."
Why AI changes the equation
Copying isn't new. What's new is the speed and the laundering.
- What used to take months of reverse-engineering now takes days.
- Output passes through a probabilistic model, losing the source's "fingerprint" and looking original.
- Proving the transfer gets harder, because no single line necessarily matches another.
The cost of stealing collapses, while the cost of building the original stays what it was: years.
The real damage
The loss isn't only revenue. The graver harm is erasing attribution — stripping the maker's name from the work, and teaching a generation of students that long-form building isn't rewarded, and that fast copying — with the machine's help — is the shortcut.
When that becomes the norm, everyone loses: fewer people dare to build something that lasts.
The way forward
I'm not arguing to stop AI — I use it daily and build with it. I'm arguing for rules that match its power:
- Code watermarking that lets origin be traced even after rephrasing.
- Transparency laws requiring disclosure of AI's use in producing a work.
- A professional norm that makes attribution the default, not the exception.
Takeaway
AI is a probabilistic engine, not a truth machine — a neutral tool in our hands. The problem isn't AI; it's letting it launder what shouldn't be laundered. If we want to reward those who build, we must protect attribution before we celebrate speed.