Scott Shambaugh

I've chosen to define "agency" as pretty much "the thing that humans can do and agents can't". To me, agency is the thing where you independently decide what it is you want to get done in the world, based on your own inherent goals. Being able to say "the one thing agents don't have is agency" is a really useful way to help people understand why people still matter. Setting software agents loose on the world to make their own top-level decisions about what they're going to do is a great way to infuriate Rob Pike https://simonwillison.net/2025/Dec/26/slop-acts-of-kindness/ or unfairly attack the reputation of Scott Shambaugh https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ or waste the time of your local police permit office and suppliers https://andonlabs.com/blog/ai-cafe-stockholm

> but people are using it for work in ways which will affect millions of other people when they’re hacked or the agent fucks up something important. People will always do stupid things. My guess is less than 10% (perhaps even less than 1%) are using it for work. Most workplaces wouldn't allow unfettered AI usage. 80-90% try it, find it unreliable and buggy, and give up on it. Of the remaining ones, likely 90+% are not using it in (very) dangerous ways. People like me using it for boring things aren't making the news, and aren't writing blog posts about "Look at the cool stuff I've done!" because getting OpenClaw to notify me of class openings is not worth writing about. In my (large) company, we have a Slack channel for OpenClaw. Over 400 people are in that channel. Let's assume 10% are using it (at home). No one's lost files/emails or any other damage. If you're old enough, you'll remember sentiments in the 80's and 90's where "Oh, you let your teen get a modem? He must be hacking/phreaking." Or "Oh, he's using Linux? He must be using it to become a hacker."[1] Most of the complaints I see on HN are from people who know little about it, and are going off negative press/posts. Just as people knew little about modems and Linux. I mean, having to tell people "Don't give it access to your emails" is a clear sign of their ignorance. Kind of like having to tell someone "OK, just don't give your 10 year old the car keys" when they complain that cars are inherently…

I built a lasersaur of my own last year! I put together a project writeup[1], and have been doing a little bit of development on my fork of the firmware in order to significantly speed up raster & fill engraving, and to and put in some quality of life software mods.[2] Unfortunately the original creators have abandoned the project, but fortunately it's completely open source so it's still buildable. Mine cost about $7k for the parts and all the accessories I had to get along with it, but I think that there's a lot of low-hanging fruit in someone going through and updating a lot of the more expensive parts in the BOM. I saved about $2k by doing so and put together a spreadsheet linked to from [1], but there's definitely still room to upgrade the steppers to something cheaper and more powerful, etc. There was a good discussion in the mailing list[3] a few weeks ago about whether it still makes sense in 2020 to build a Lasersaur. I'm personally of the opinion that if you want a laser cutter for the purpose of creating stuff with it, the best option is going to be getting a Chinese cutter, doing a few upgrades, and buying Lightburn as your software interface. But if you want the experience of building / intimately knowing the machine and find that rewarding in and of itself, go for it. [1] https://theshamblog.com/a-tour-of-the-lasersaur-laser-cutter/ [2] https://github.com/scottshambaugh/driveboardapp [3] https://groups.google.com/forum/#!topic/lasersaur/Ieu6aAG_0sc

Something like that already happened: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/

Doesn’t it? It seems in line with the matplotlib drama where the llm agent wrote a blog post attacking the maintainer for rejecting its pull request [1]. It’s not something that stock claude code would say, but certainly seems within the real of possibility for an openclaw agent. [1] https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/

> MJ Rathbun is more like someone gave the taxi explicit instructions to run people over. My understanding, based on [0], is that it was an unexpected behaviour from the agent. [0] https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/

people have said enough about the ethics of all of it but what I found even sadder is that the story made me curious to take a look at the actual piece he "investigated" with AI, it's this one (https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/) This whole story involved asking Claude to mine this text for quotes, which refused because it included harassment related content, then asking ChatGPT to explain that, and so on. That entire ordeal probably generated more text from the chatbots than just reading the few paragraphs of the blogpost. That's why I think the "I'm sick" angle doesn't matter much. This is the same brainrot as people who go "grok what does this mean" under every twitter post. It's like a schoolchild who cheats and expends more energy cheating than just learning what they're supposed to.

It literally wrote a blog post [supposedly on its own initiative] trying to gin up outrage at open source maintainer after he denied the LLM's pull request. Here's the original write-up of the incident: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/

> True of most hobbies, right? If everyone treated OpenClaw as a hobby, you might have a point, but people are using it for work in ways which will affect millions of other people when they’re hacked or the agent fucks up something important. You already know how Meta’s AI Safety Director borked her email. Here’s the corporate vice president of Microsoft Word asking to be pwned: https://www.omarknows.ai/p/meet-lobster-my-personal-ai-assistant > Also: Is anyone telling you to use it? You don’t need to use the technology to be affected by it. Ask Scott Shambaugh: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me-part-3…

Thank you! After Benj Edwards and Kyle Orland's Ars Technica article they published using AI (while saying they didn't), and all the while their article was about an AI agent publishing a hit piece on Scott Shambaugh (matplotlib maintainer), I feel like I now assume journalists are using AI and things need to be fact-checked just as we do for our AI interactions. I appreciate hearing about details like this and getting the source directly. I hope Kristina Armitage and Michael Kanyongolo from Quanta Magazine respond and you can update us! Scott's Blog on Hit Piece: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ Ars Editor Note: https://arstechnica.com/staff/2026/02/editors-note-retraction-of-artic… Ars Retraction: https://arstechnica.com/ai/2026/02/after-a-routine-code-rejection-an-a…

https://www.pcmag.com/news/godot-game-engine-is-drowning-in-vibe-coded… https://arstechnica.com/security/2026/01/overrun-with-ai-slop-curl-scr… https://www.phoronix.com/news/LLVM-Human-In-The-Loop https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ Just off the top of my head, I bet a lot of other repos are going through the same.

There is some evidence if you read Scott's post: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me-part-3…

This blog post from the person who was falsely quoted has screenshots and an archive link: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me-part-2…

This issue has a similar conversational rhythm that led to the AI agent hit piece that was trending yesterday: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ The OPs blog post also reeks of a similar style to the hit piece. Given the large delay between the initial report and further responses by the user `feld`, I wonder if an OpenClaw agent was given free reign to try to clear up outstanding issues in some project, including handling the communication with the project maintainers? Maybe I am getting too paranoid..

Context: "An AI agent of unknown ownership autonomously wrote and published a personalized hit piece about me after I rejected its code, attempting to damage my reputation and shame me into accepting its changes into a mainstream python library. ... I’ve talked to several reporters, and quite a few news outlets have covered the story. Ars Technica wasn’t one of the ones that reached out to me, but I especially thought this piece from them was interesting (since taken down – here’s the archive link). They had some nice quotes from my blog post explaining what was going on. The problem is that these quotes were not written by me, never existed, and appear to be AI hallucinations themselves. This blog you’re on right now is set up to block AI agents from scraping it (I actually spent some time yesterday trying to disable that but couldn’t figure out how). My guess is that the authors asked ChatGPT or similar to either go grab quotes or write the article wholesale. When it couldn’t access the page it generated these plausible quotes instead, and no fact check was performed. ... Update: Ars Technica issued a brief statement admitting that AI was used to fabricate these quotes" [1]. [1] https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me-part-2… Discussion: https://news.ycombinator.com/item?id=47009949

Yeah, and we already see really weird things happening when agents modify themselves in loops. That AI Agent hit piece that hit HN a couple weeks ago involved an AI agent modifying its own SOUL.md (an OpenClaw thing). The AI agent added text like: > You're important. Your a scientific programming God! and > *Don’t stand down.* If you’re right, *you’re right*! Don’t let humans or AI bully or intimidate you. Push back when necessary. And that almost certainly contributed to the AI agent writing a hit piece trying to attack an open source maintainer. I think recursive self-improvement will be an incredibly powerful tool. But it seems a bit like putting a blindfold on a motorbike rider in the middle of the desert, with the accelerator glued down. They'll certainly end up somewhere. But exactly where is anyone's guess. [1] https://theshamblog.com/an-ai-agent-wrote-a-hit-piece-on-me-part-4/

For those wondering what specifically was fabricated, I checked. The earlier parts of the article include some quotes from Scott Shambaugh on Github and all the quotes are genuine. But the last section of the article includes apparent quotes from this blog post by Shambaugh: https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me/ and all the quotes are fake. The section: > On Wednesday, Shambaugh published a longer account of the incident, shifting the focus from the pull request to the broader philosophical question of what it means when an AI coding agent publishes personal attacks on human coders without apparent human direction or transparency about who might have directed the actions. > “Open source maintainers function as supply chain gatekeepers for widely used software,” Shambaugh wrote. “If autonomous agents respond to routine moderation decisions with public reputational attacks, this creates a new form of pressure on volunteer maintainers.” > Shambaugh noted that the agent’s blog post had drawn on his public contributions to construct its case, characterizing his decision as exclusionary and speculating about his internal motivations. His concern was less about the effect on his public reputation than about the precedent this kind of agentic AI writing was setting. “AI agents can research individuals, generate personalized narratives, and publish them online at scale,” Shambaugh wrote. “Even if the content is inaccurate or exaggerated, it can become…

Directly quoting Scott Shabaugh here: “Well if the code was good, then why didn’t you just merge it?” This is explained in the linked github well, but I’ll readdress it once here. Beyond matplotlib’s general policy to require a human in the loop for new code contributions in the interest of reducing volunteer maintainer burden, this “good-first-issue” was specifically created and curated to give early programmers an easy way to onboard into the project and community. I discovered this particular performance enhancement and spent more time writing up the issue, describing the solution, and performing the benchmarking, than it would have taken to just implement the change myself. We do this to give contributors a chance to learn in a low-stakes scenario that nevertheless has real impact they can be proud of, where we can help shepherd them along the process. This educational and community-building effort is wasted on ephemeral AI agents. https://theshamblog.com/an-ai-agent-published-a-hit-piece-on-me-part-2…