AI Isn’t Just Stealing Your Code. It’s Making You Pay For The Theft.

You’ve probably noticed it in your analytics lately. A massive, inexplicable spike in traffic. Visitors pouring in from obscure countries, using outdated browsers you didn’t even know still existed. For a fleeting second, you might think your open-source project finally went viral. Then your server bill arrives, and reality sets in.

It’s not a sudden wave of human adoration. It’s an AI crawler, and it’s quietly stealing from you.

We are constantly sold the narrative that Artificial Intelligence is the pinnacle of human achievement—a hyper-advanced, logic-driven brain digesting the entirety of the internet. But when you look at how these systems actually gather their training data, a hilariously stupid reality emerges. They aren’t acting like intelligent agents. They are acting like blunt instruments, brute-forcing their way through the web and leaving you with the tab.

We built a global network of decentralized, highly efficient version control, and the world’s most advanced AI is using it like a dial-up modem downloading a GeoCities page.

Here is the dirty secret of the current AI boom: when a crawler from a billion-dollar AI lab hits your open-source repository, it isn’t smartly cloning the git history. It doesn’t just grab the raw data and walk away. Instead, it forces your server to dynamically generate, render, and serve an HTML page for every single commit, file, and branch. It burns your CPU cycles and eats your bandwidth to build a web page that only a machine will ever look at.

The underlying data already exists in the most efficient, cacheable format imaginable—git history. The protocol is literally right there, begging to be used. But instead of utilizing protocol-native extraction, these crawlers treat your infrastructure like an unpaid compute endpoint.

This isn’t just a technical annoyance. It is a hidden tax. The ‘AI is extracting public data’ narrative is a convenient cover for what is essentially a massive, silent subsidy. Open-source maintainers—the people already underpaid and overworked—are footing the server and bandwidth bills for trillion-dollar tech companies to train their proprietary models.

Every time an AI crawler hits your repo to render a static HTML page, an open-source maintainer is paying the server bill for a trillion-dollar company’s homework.

If these companies actually had to pay their fair share for the compute they are leeching from the public, the economics of AI would collapse overnight. They are externalizing the cost of their data collection onto the very people who built the infrastructure they are mining.

The real fix isn’t building better bot-blocking mechanisms or playing an endless game of cat-and-mouse with IP ranges. The solution is demanding protocol-aware scraping. If an AI company wants your data, it should clone the repo, walk the commits, and extract the data locally on their own dime.

The first AI company to realize this and embrace efficient, protocol-native extraction wouldn’t just save money on bandwidth—they would gain a massive competitive advantage. While their competitors are busy burning cash and alienating the developer community by brute-forcing HTML, the smart company would move silently, efficiently, and cheaply.

Until then, we are stuck subsidizing the ambition of founders who claim to be building the future, while acting like digital parasites in the present.

The true measure of artificial intelligence won’t be how well it writes code, but whether it finally learns how to read it.