Warning: This post is more of a rant that I needed to dump somewhere, it’s not very insightful.

A particular phrase that I hate being told goes something like this:

Every big open source project uses AI code, get used to it.

I was thinking about this recently when trying to figure out how to block AI scrapers and agents from Hacks Guide Wiki, in a similar vein to The Cutting Room Floor’s recent attempts. I was suggested to use Anubis, which provides a proof-of-work challenge to visitors to block scraping. But Anubis itself accepts AI code, as seen in its AGENTS.md.¹ That makes it unappealing to me.

But wait, MediaWIki itself accepts generative AI code, and I’m using that to host the wiki. There are even bigger projects like the Linux kernel that also accept AI code, and most of my own devices and servers run Linux. So if I am already running software that was made with the help of AI, should I just give up pushing back?

Well, no. I don’t think it’s all or nothing. I have a strong distaste for generative AI, but I also am very busy in my life and I have to pick my battles. I know Linux and MediaWiki accept them, but neither are immediately practical to replace. They’re both big projects that aren’t easy to fork and maintain, and you likely would not want to run older versions due to known security vulnerabilities.

This very blog (as of August 2026) is hosted on WordPress.com, a service and application that itself promotes the use of AI. The main reason I have not moved yet is because it’s low-priority and not a trivial process, not because I think it is fine.

But I can reject other instances of AI code that I am not forced to deal with. I have an older version of rsync installed on my PC, before the use of Claude, because I don’t trust newer versions after important features broke. I moved some of my active projects to Codeberg because I don’t like how much GitHub is promoting usage of Copilot. I disabled pull requests on most of my repositories in part because I am uncomfortable with someone attempting to put AI-generated code into my projects.

Generative AI is not the future, it’s not inevitable, it’s not here to stay. I hate being told to accept it. I hate how it has ruined technology, raising prices of hardware and reducing software quality. I hate how we’re expected to outsource our thinking to the lying theft machine.

Sidenote: AI ports and recompilation projects are worse than useless

I have seen a few recompilation projects, such as the Animal Crossing GCN PC port and Mario Kart Wiicompiled, state that they were made with the assistance of AI tools like Claude. And like, why? Why would you do that? You don’t want to do it properly, you just want to outsource your thinking to get something that technically works but is difficult to maintain and develop?

I was originally excited when I heard these came out, but then became completely uninterested when I learned that AI was used at any point. Sorry, I don’t give a shit about these slop ports anymore.

But since these projects exist, now it’s even less likely someone will attempt to make a proper port of these decompilation projects. These slop projects are worse than useless. Great job!

Fun fact, my post about it got mentioned in a Time Extension article.

Sidenote: Fuck Claude

Recently Xkeeper posted about self-hosting in the era of LLMs. There is a particular section about Claude and how it actively attempts to evade bans by switching between User-Agents. Since then, The Cutting Room Floor now bans Claude users. It also serves a page intended to mess with AI chatbots.

Claude doing this makes me really upset. These companies and their users do not respect consent when being told to fuck off. Claude is actively malicious to web hosts.

Footnotes

1. Someone will probably mention “this is to mess with agents that don’t follow the rules!” which means it isn’t a rejection of AI code. There are merged pull requests that openly state they were assisted by certain LLMs. ↩︎