Anthropic Panic-Deleted Thousands of GitHub Repos Like a Crypto Bro

Thousands of GitHub repos got nuked because Anthropic’s source code allegedly “oopsie-daisy” leaked onto the internet and somebody finally remembered the door has a lock.

This is the part where every Open Source Guy who owns three mechanical keyboards starts screaming about “transparency,” and every AI company starts screaming about “IP,” and somehow they’re both holding the same knife.

Anthropic filed takedowns across GitHub to remove repositories that contained the leaked code. They called it an accident.

And yes, the irony is so thick you could spread it on toast: these companies train models on the public internet like it’s an all-you-can-eat buffet, then act personally victimized when their own recipes end up on the sidewalk.

The open-source crowd wants auditable AI because locked models can hide bias, abuse, and “whoops, it told a teenager how to make napalm.” The AI companies want locked models because that’s how you charge rent on math.

Meanwhile, this is what “AI safety” looks like in 2026: not guardrails, not ethics, not alignment — just DMCA paperwork and a frantic attempt to stuff toothpaste back into the tube.

If you write code for a living, this isn’t drama. It’s the business model. Your work gets vacuumed up for training data, then the companies that vacuumed it up tell you their code is sacred scripture.