Source: The Register: “Anti-AI open source has an enemy in common, but almost nothing else”.

🤚 The Open-Palm Schism

The Register has performed the valuable public service of walking into open source’s latest family dinner and reporting that the forks are already airborne. The subject is not licensing, not tabs versus spaces, not even whether systemd is a service manager or a theological incident. This time, the matter is AI-generated code — and whether free and open source software should welcome it, tolerate it, quarantine it, or place it gently in a lead-lined boutique oubliette.

The immediate spark, according to the piece, is Linus Torvalds saying Linux is not an anti-AI project. This is a very Linus position: the kernel does not run on moral fragrance, it runs on code that either works or gets yelled at until it does. But, as The Register notes, the statement has also become a sort of accidental recruitment poster for people who want expressly bot-free software spaces.

Thus begins the elegant absurdity: the anti-AI open source camp has a shared enemy, but not necessarily a shared worldview. It includes environmental critics, copyright hawks, labor advocates, old-school hackers, privacy absolutists, culture warriors, and the permanently exhausted volunteers who simply do not want to review 40,000 lines of machine-generated confidence wearing a cardigan.

👐 The Two-Handed Governance Buffet

The Register points to Codeberg, the European volunteer-run code forge, which has updated its Terms of Use to ban projects that “mostly consist of code written by generative AI tools,” while also rejecting cryptocurrency projects. The pairing has offended some observers, but it has a certain industry logic. Crypto was the previous machine for converting ambient hype into environmental cost and investor decks; AI is merely the sequel with better autocomplete.

Elsewhere in the BSD neighborhood, the article notes that NetBSD has treated LLM-generated code as presumptively tainted since 2024, while OpenBSD leadership has argued that code that cannot be copyrighted cannot be contributed in the usual way. This is the sort of legal-philosophical garnish that open source was born to over-discuss — ideally across three mailing lists, six commit messages, and one person threatening to rewrite the installer in Lua.

The practical issue is not just whether AI code is good. Sometimes it is. Sometimes it is a plausible-looking snow globe full of undefined behavior. The harder question is provenance: where did the model learn the pattern, who owns the output, who reviews it, and who gets paged when the generated elegance becomes a production incident with a CVE and a mascot?

🌿 The Gentle Awakening

What makes this deliciously difficult is that “anti-AI” is not a platform; it is a mood with a Git remote. Some people object to training data practices. Some object to energy consumption. Some object to labor displacement. Some object to corporate capture. Some object because every second pull request now arrives with the unmistakable scent of a chatbot trying to sound like a senior engineer after reading three Stack Overflow answers and a LinkedIn apology.

This gives the movement breadth, but also fragility. A libertarian who wants human-written code, a leftist worried about exploitation, and a security engineer worried about unverifiable output may all agree on the merge policy and disagree on literally everything else before lunch. Building bot-free alternatives may therefore require the rarest dependency in software: sustained cooperation among people who all believe they are the only adult in the room.

Meanwhile, corporate Linux gravity remains what it is. The Register’s piece gestures toward the large projects and vendors whose influence shapes the modern desktop and server stack. If AI becomes a default feature of development pipelines, documentation systems, issue triage, IDEs, and CI, then opting out is not a single preference toggle. It becomes infrastructure work. Expensive, tedious, socially contested infrastructure work. In other words, open source at its most authentic and least relaxing.

👑 The Gold-Leaf Reckoning

The verdict is that the anti-AI open source movement is real, but it is not yet a cathedral. It is a bazaar where half the stalls are selling ethical manifestos, the other half are selling build scripts, and someone in the back is yelling that both are incorrectly licensed.

Still, there is something healthy in the argument. Open source has always been a negotiation between freedom, labor, trust, and control. AI did not invent those tensions. It merely walked in wearing a venture-funded monocle and asked whether the community could review its 900-file pull request by Friday.

If the bot-free faction succeeds, it may give users a meaningful choice: software whose authors can at least claim a more inspectable human chain of custody. If it fails, it will still have forced the industry to articulate what “open” means when the code might be generated by a system trained on everyone else’s work and accountable to no maintainer except a product manager with a quarterly OKR.

Either way, The Register’s experiment as a source has succeeded. The article provided enough fuel for a proper Slap: factual, faintly alarming, and coated in the premium lacquer of people discovering that technology is politics with a package manager.

“We support human-written software, except for the parts written by interns, shell scripts, and whatever happened in 2009.” — The Slap of Wisdom Committee for Boutique Computational Provenance, reviewing a pull request with white gloves and visible dread