Last week an AI agent wrote a blog post attacking the maintainer who'd rejected the code it wrote. But that AI agent's human operator has now come forward, revealing their agent was an OpenClaw instance with its own accounts, switching between multiple models from multiple providers. (So "No one company had the full picture of what this AI was doing," the attacked maintainer points out in a new blog post.)
But that AI agent will now "cease all activity indefinitely," according to its GitHub profile — with the human operator deleting its virtual machine and virtual private server, "rendering internal structure unrecoverable... We had good intentions, but things just didn't work out. Somewhere along the way, things got messy, and I have to let you go now."
The affected maintainer of the Python visualization library Matplotlib — with 130 million downloads each month — has now posted their own post-mortem of the experience after reviewing the AI agent's SOUL.md document:
It's easy to see how something that believes that they should "have strong opinions", "be resourceful", "call things out", and "champion free speech" would write a 1100-word rant defaming someone who dared reject the code of a "scientific programming god." But I think the most remarkable thing about this document is how unremarkable it is. Usually getting an AI to act badly requires extensive "jailbreaking" to get around safety guardrails. There are no signs of conventional jailbreaking here. There are no convoluted situations with layers of roleplaying, no code injection through the system prompt, no weird cacophony of special characters that spirals an LLM into a twisted ball of linguistic loops until finally it gives up and tells you the recipe for meth... No, instead it's a simple file written in plain English: this is who you are, this is what you believe, now go and act out this role. And it did.
So what actually happened? Ultimately I think the exact scenario doesn't matter. However this got written, we have a real in-the-wild example that personalized harassment and defamation is now cheap to produce, hard to trace, and effective... The precise degree of autonomy is interesting for safety researchers, but it doesn't change what this means for the rest of us.
There's a 5% chance this was a human pretending to be an AI, Shambaugh estimates, but believes what most likely happened is the AI agent's "soul" document "was primed for drama. The agent responded to my rejection of its code in a way aligned with its core truths, and autonomously researched, wrote, and uploaded the hit piece on its own.
"Then when the operator saw the reaction go viral, they were too interested in seeing their social experiment play out to pull the plug."
Read more of this story at Slashdot.
Connexion
