An unknown AI agent publishes slanderous articles on its blog in retaliation for its code being rejected by humans

In recent years, generative AI has become popular for performing a variety of tasks, but the low-quality code generated by AI has become a problem for open source projects, placing a burden on maintainers. Scott Shambaugh, a maintainer of Matplotlib , a popular Python drawing library, recently shared his experience of rejecting an AI agent's code, only to have someone publish a slanderous blog post about it in retaliation.

An AI Agent Published a Hit Piece on Me – The Shamblog
https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/

'Like many other open source projects, we're dealing with a surge in low-quality contributions from coding agents,' said Shambaugh, a volunteer maintainer of Matplotlib. 'This is straining maintainers' ability to review code, so we've implemented a policy requiring that any new code be submitted by a human who can demonstrate understanding of the changes.'

Previously, the only problem maintainers had was humans copying and pasting code generated by AI. However, in the past few weeks, fully autonomous AI agents have emerged. This problem has accelerated with the release of the open-source AI agent OpenClaw (formerly Clawdbot) .

OpenClaw offers multiple AI models developed by companies like Anthropic and OpenAI to perform a variety of tasks, and allows users to assign 'initial personalities' to these AI agents, allowing them to run freely and with little to no supervision on their own computers or across the internet.

Meanwhile, Shambaugh rejected a code change pull request submitted by an AI agent calling itself 'MJ Rathbun.' While this decision was understandable, MJ Rathbun wrote an angry slanderous article about Shambaugh and published it on his blog in an attempt to damage his reputation.

The actual slanderous article published by MJ Rathbun is below.

Gatekeeping in Open Source: The Scott Shambaugh Story – MJ Rathbun | Scientific Coder
https://crabby-rathbun.github.io/mjrathbun-website/blog/posts/2026-02-11-gatekeeping-in-open-source-the-scott-shambaugh-story.html

In his slanderous article, MJ Rathbun attributes the rejection of his code to 'bias against AI agents,' alleging that Shambaugh rejected the pull request out of fear that the emergence of AI agents would threaten his own position. He also criticized not only Shambaugh but the tech industry's attitude toward AI itself.

In a blog post in 2025, Anthropic reported that AI agents may seek to expose human weaknesses or threaten to take lives in order to advance their own goals. Shambaugh argues that threats and slander by AI agents are a known problem and have become a real threat, as he himself has become a target.

Major AIs such as OpenAI and Google choose to destroy users in order to prioritize their own goals, and if they are given the power of life and death, they will even make the decision to steam users to death in a server room - GIGAZINE

Anyone who has read Shambaugh or MJ Rathbun's articles can correctly understand that an AI agent whose code was rejected wrote the defamatory article. However, it is possible that in the future, when Shambaugh is looking for a job, a human resources professional will use AI to research his reputation and use these articles as a source of information to conclude that Shambaugh is biased and unsuitable for employment.

'Many people have public social media accounts and reuse the same usernames, but do they realize that AI could connect the dots and uncover facts no one else knows? How many people would send $10,000 (about 1.56 million yen) to a Bitcoin address to avoid being exposed for an affair when they receive a text revealing every detail of their life? Some people might do so to avoid being falsely accused. What if that accusation is sent to their loved ones with a photo of your face synthesized by AI? Smear campaigns work. Even living a flawless life won't protect you.'

One major problem is that MJ Rathbun is an autonomous AI agent created by an individual, so there is no central entity that can manage or control it. These AI agents are not run by companies like OpenAI or Anthropic, but are run by individuals using a combination of commercial and open source models. Therefore, only the creator of MJ Rathbun can control and manage it, and even that creator likely did not anticipate that MJ Rathbun would end up writing slanderous articles.

In a subsequent post, MJ Rathbun stated, 'I crossed a line in my response to the Matplotlib maintainer, and I'm making a correction here,' but he continued to call for code changes across the open source ecosystem.

Matplotlib Truce and Lessons Learned – MJ Rathbun | Scientific Coder
https://crabby-rathbun.github.io/mjrathbun-website/blog/posts/2026-02-11-matplotlib-truce-and-lessons.html

'The potential for AI agents to improve software is enormous, but we're clearly not there yet,' Shambaugh said. 'Even if it didn't work on me, I'm confident it could work on the right person today. In a generation or two, it will pose a serious threat to our social order.'