If you’ve been following the AI + open source community for the past week, you’ve probably heard some rumblings about Moltbot (formerly known as Clawdbot), and how it can be deployed to autonomously do tasks for you. Some of those tasks can include automatic email handling, calendaring, etc. But the most know use case right now is the unfortunate chaos that is has unleashed on open source projects.
It all started with a pull request to matplotlib, where the PR was closed by scottshambaug because it was created by an AI agent and per previous discussions, they would not allow changes be made using just AI agents.
That unfortunately took a wrong turn, when the agent then posted a defamatory article on Scott, apparently all autonomously, and shared it as link. The blog post has since been taken down, but its content is still available here.
This then led Scott to post a response blog called “An AI Agent Published a Hit Piece on Me”, with followups in part 2 and part 3.
Scott’s blogs and the ensuing press surrounding this open source drama led to the AI agent’s operator (or is it really?) to publish their response, and I have to say, it’s terribly written. It’s not an apology for the headache they’ve caused, merely a “it wasn’t me, I just unleashed the agent without supervision”. Finally, as of yesterday, the agent posted more information about itself and has now shut down its operations, per the GitHub account it controls.
This has been a wild ride for anyone interested in open source and AI, and for myself, as I use and participate in both communities, it’s frustrating to see that the power of AI can be used for such annoying and frustrating means. AI can be used for good, but requires supervision. It is not mature enough to just be let loose on other people’s codebases, and assume that they willingly will accept the changes being made. This is abhorrent behavior from the real Rathbun operator. They could’ve used this as a great tool, but instead they just wanted some quick wins and used AI to produce garbage that impacted real humans. I’m glad to see this gets the attention it needs, but I’m sorry it happened to Scott, and I’m sure others will probably impacted in a similar way within the near future.
Please use AI responsibly!