AI companies fear data leaks, but end up leaking themselves first.

The incident currently spreading wildly on X and GitHub feels like a scripted scenario, perfectly illustrating the fragility of the entire AI industry.

Here's what happened:

— Anthropic accidentally released an update on npm
— The package contained a roughly 60MB debug file
— Inside was a snippet of Claude's internal code

And then the show began.

23 minutes later, a researcher (Chaofan Shou) discovered it, downloaded it, and made it public.
Within hours: millions of views.

By the time the Anthropic team realized, the code had already:
— Gone viral on GitHub
— Been forked tens of thousands of times
— Started to take on a life of its own

A classic Web2 mistake…
Happening in an era where information spreads like a virus.

Now, pay attention to the second part.

A developer appeared: Sigrid Jin.

He:
— Analyzed the leaked code
— Rewrote it into Python in 8 hours
— Gained tens of thousands of stars on GitHub
— Then rewrote it again in Rust

Done.
This product essentially became an open-source clone.

But the real point is here.

Within the code, they discovered a system called the “Undercover Mode.”

Its task was:
To prevent internal data leaks from the model.

In other words:

A company specializing in leak prevention…
Leaked its own code through a debug file.

This isn’t just irony.
It’s a signal.

What does it mean?

1. AI companies aren’t as “closed” as they seem
Any mistake → the backend gets exposed immediately.
2. The moat around AI is much more fragile than investors think
If a product can be quickly copied → value shifts:
— From the model → to the data
— From the data → to distribution
3. The speed of copying = a new reality
What used to take years,
Now takes only hours.

What are the consequences?

— Profit margins for AI companies are shrinking
— The open-source movement accelerates
— “Secret technologies” depreciate

And most importantly:

The market is beginning to realize that
value isn’t in the code itself, but in the ecosystem and user reach.

We are moving toward a world where:

— Models become commodities
— The winners aren’t the “most powerful AI” companies
— But those with the strongest distribution, products, and funding

Today’s AI reminds people of the early days of the internet.
Everyone thought technology would win.

But in the end,
it’s those who control attention who come out on top.