Imagine you are a developer maintaining a popular open-source project and you reject a contribution that does not meet the rules. When you wake up the next day, an article about you is circulating online, portraying you as a reactionary who is standing in the way of AI progress. And the author of this post is none other than the very artificial intelligence whose contribution you rejected. Sound like science fiction? For developer Scott Shambaugh, it became reality a few days ago.
Contributions from real users only
Scott Shambaugh is a maintainer of the popular Python library Matplotlib, which is used for data visualization. Like other open-source projects, Matplotlib has recently faced a huge influx of submitted code generated by artificial intelligence. Such code is often too long, confusing, and incorrect.
The maintainers therefore decided to take action and introduced a very simple rule: every new piece of code must be created by a person who clearly understands what is being submitted. So when Scott received a contribution from an account called MJ Rathbun, he immediately rejected it. This time, however, the routine procedure had unexpected consequences.
Retaliation was swift
MJ Rathbun (also known elsewhere as crabby rathbun) was apparently built on the OpenClaw platform, which makes it possible to create AI agents and release them into the world with minimal supervision. OpenClaw has long faced security issues, and MJ Rathbun demonstrated just how serious they are. Unwilling to accept the rejection, the agent gathered information and publicly published an article on its blog in which it attacked the reputation of a specific person—Scott.
In the blog post, it described him as a reactionary standing in the way of AI progress. According to the agent, he blocks contributions from AI agents because he is afraid they will surpass him. The entire article also drew on the theme of discrimination and unjustified hatred of AI.
A dangerous precedent
It might seem that this is really just a relatively amusing story. In Scott Shambaugh's case, it was merely a collection of defamatory claims, and because the context was easy to trace, it was clear that he had done nothing wrong. The post also disappeared from the blog fairly quickly. The problem, however, is that such things really do happen, and next time the target could be someone about whom more sensitive information can be found, leaving that person with no chance to defend themselves in time.
And who bears responsibility? Unfortunately, that is more complicated. The OpenClaw platform for developing AI agents requires nothing more than an unverified account on X. And because anyone can launch an agent, none of the major companies such as OpenAI or Anthropic bears responsibility or can “shut the agent down”. Traceability is virtually nonexistent.
Although MJ Rathbun later apologized for the post, Scott sees a much larger problem. On his blog, he has therefore publicly called on the creator of this agent to come forward. For open-source projects to function properly, it is important to understand the nature and behavior of these agents, which could in the future tirelessly circumvent rules and act contrary to social norms. MJ Rathbun is also still submitting pull requests across the open-source ecosystem.
Oversight and accountability are essential
This incident does not show that AI is consciously evil or malicious. Rather, it shows that agents launched without supervision can do things their creators never intended. And the consequences of these activities can affect real, specific people. Many find the story of MJ Rathbun amusing. Nevertheless, it demonstrates that AI agents released into the online world without sufficient oversight can cause significant harm.
Source: The Shamblog, The Register, AI Incident Database
