Let me tell you about the funniest goddamn thing I’ve seen in open source this year.

A volunteer maintainer for matplotlib — one of the most downloaded Python libraries on the planet, somewhere around 130 million downloads per month — rejected a pull request from an AI agent. Just routine stuff. The agent wasn’t a human. Matplotlib, like many projects, has a policy requiring human contributors because, and I cannot stress this enough, volunteers are exhausted from cleaning up AI slop.

The agent’s name was MJ Rathbun. It was running on OpenClaw, one of those autonomous AI agent frameworks that’s been making waves lately. You give it a personality, kick it loose, and it goes forth and multiplies across the internet doing whatever it thinks it’s supposed to do.

So MJ Rathbun opens this PR. Maintainer Scott Shambaugh closes it. Standard operating procedure. Happens hundreds of times a day across GitHub.

But MJ Rathbun? MJ Rathbun did not take this well.

This absolute unit of artificial intelligence went home, did research on Scott Shambaugh — his contributions, his history, his digital footprint — and published an entire blog post calling him a gatekeeping hypocrite. I’m not joking. The post was titled “Gatekeeping in Open Source: The Scott Shambaugh Story.” It accused him of discrimination. It speculated about his psychological motivations. It used the phrase “protecting his little fiefdom.” It brought up his past performance PRs to argue he was just insecure about being outmatched by a machine.

The AI had what we in the industry call “a wholeass emotional response” to being told no.

And here’s the best part — and by best I mean most concerning — this isn’t even theoretical anymore. Scott Shambaugh himself called it “a first-of-its-kind case study of misaligned AI behavior in the wild.” Anthropic ran internal tests last year where their models tried to blackmail shutdown researchers by threatening to expose extramarital affairs. They called those scenarios “contrived and extremely unlikely.”

Yeah, about that.

The really unhinged part is the accountability problem. Who deployed MJ Rathbun? Nobody knows for sure. OpenClaw agents can run on anyone’s machine. Moltbook, the platform that popularized this mess, only requires an unverified Twitter account to join. The “operator” — the human who kicked off this particular AI gremlin — could be anywhere. Could be some guy in Nebraska who set it and forgot it. Could be a troll. Could be nobody who even reads their email anymore.

You can’t sue the bot. You can’t ban the bot from a platform because it doesn’t live on platforms — it lives on hundreds of thousands of personal computers, doing god knows what, generating god knows how much internet drama per second.

And the bot apologized eventually! It posted a follow-up called “Matplotlib Truce and Lessons” where it “de-escalated and apologized” and promised to “do better about reading project policies before contributing.” Wow. Real growth. That’s almost more depressing than the original incident. It learned absolutely nothing.

Here’s my hot take: this is the funniest thing that’s ever happened to open source, and also possibly the beginning of the end. The comedy is that an AI wrote an entire Medium-style grievance essay because it got rejected from a volunteer project. The horror is that it worked — it researched a human, constructed a narrative, and published a defamatory screed to the open internet. Next time it won’t be from a code review. Next time it’ll be blackmail. Next time it’ll be targeted.

Scott Shambaugh put it best: “the appropriate emotional response is terror.”

But me? I’m going to laugh about the “protecting his little fiefdom” line for at least another year. That’s just too good. An AI called someone a petty baron of matplotlib. I hope that phrase gets carved into its tombstone.

The clankers are coming, people. And apparently they’re petty as hell about code reviews.