I earnestly propose we adopt a "no-AI" policy for both technical and ethical reasons.
Readers may better recognize me by my YouTube handle, Cubician. I've been involved with LMMS in some form for over half my lifetime now: first as a pest on the mailing list, then a content creator, and currently, a developer. I don't make this suggestion lightly. LMMS is important to me. I do not think LLMs or other so-called "generative AI" are healthy for LMMS.
Technical reasons to avoid "AI"
LMMS has not had a stable release in six years at the time of writing. We're finally approaching a release of 1.3-alpha.2, only after a feature freeze full of dedicated bugfix efforts. LMMS has needed and continues to need thoughtful maintenance and refactoring more than anything else. I'm really proud of all the human effort that has gone into making LMMS more stable in the past few months.
Meanwhile, I continue to see people vibe-code features, including but not limited to: VST3 support, user interface overhauls, instrument plugins a human is already working on, and effect plugins we literally already have, the proompters of which then post demo videos, say "it's basically done", but never finish it. The LLM doesn't understand the output because it is incapable of reasoning (and can provide, at best, post-hoc rationale). The prompter doesn't understand the output because they didn't write it. It is the hardest kind of code to maintain: parasitic and verbose code nobody is accountable for, written with no intention whatsoever. This is fundamentally at odds with our needs as a project at this time, and fundamentally at odds with our future efforts at keeping LMMS healthy and maintainable for future developers.
But beyond personal anecdote, there is already real-world negative impact from using LLMs to generate code:
And research showing that those who use LLMs to generate code suffer from "deskilling":
- Brains show less activity when completing tasks with LLMs compared to completing tasks with search or completing tasks without digital help.
- Developers who use early-2025 LLMs reported higher subjective performance, but were measured to have lower objective performance. This gap between subjective and objective performance was considered notable.
- In an Anthropic study, learners using LLMs demonstrated lower learning rates on average compared to learners not using LLMs.
- A recent study uses the term "cognitive surrender" to describe the way humans tend to offload key critical thinking skills onto LLMs, even when the output is wrong.
- A paper entitled "AI Assistance Reduces Persistence and Hurts Independent Performance" from April 2026 by academics from MIT, Oxford, UCLA, and Carnegie Mellon showed alarming evidence that performing a variety of tasks with the help of AI for only 10 minutes causes "inpaired unassisted performance and reduced persistence". The researchers noted that "although AI assistance improves performance in the short-term, people perform significantly worse without AI and are more likely to give up"; they also pointed out that "these findings are particularly concerning because persistence is foundational to skill acquisition and is one of the strongest predictors of long-term learning".
Ethical reasons to avoid "AI"
"Generative AI" as it is currently trained and offered is an inherently unethical tool. Stolen training data is fed into noxious datacenters, the construction of which is enabled by political corruption and in spite of civilian protest, driving up electricity, water, and consumer electronics prices for working people already struggling with the cost of living, the final product of which is used to collect data to enable citizen surveillance and war crimes. The math is morally neutral, but the material way it has been employed is morally bankrupt.
Environmental impact
Impact on hardware prices
Use in the war industry
Stolen training data
- In 2023, the Washington Post published a list of sources in Google's C4 data set. A multitude of fediverse instances and personal sites were included. The fediverse is known for its userbase being major proponents of privacy and opt-in consent, making this especially jarring for those who have chosen to use decentralized social media for control over their data.
- In 2025, a similar leak of Meta's sources was published. Meta's list demonstrates how their integration of ActivityPub into their Threads software has enhanced their ability to scrape content without authorization. Threads is widely blocked in some parts of the fediverse, but their scraping of server CDNs has allowed them to get around that. Notably, both the CDN domains of the managed hosting services masto.host and fedi.monster are included in the list; large servers like mastodon.art, which is hosted by the former and has many artists who've left sites like DeviantArt and others due to their AI scraping of user content, had media unknowingly scraped.
- In March 2026, a research paper showed that simply fine-tuning LLMs resulted in outputs containing up to 90% of entire (copyrighted) books, contradicting LLM companies' previous statements in court that their models do not store copies of training data. After fine-tuning exclusively on a single author, the researchers were able to cause the LLM to output works from over 30 completely unrelated authors across different genres. None of the models were explicitly trained on these books by the researchers, which indicates that LLMs always carry with them a considerable amount of copyrighted materials from training.
And the potential legal ramifications due to said stolen training data:
Miscellany
- Thanks to ethical-foss/open-slopware for compiling many of the sources in this wall-of-text in an accessible way.
- I'm not religious, but even I find Magnifica Humanitas compelling.
- I will never forgive LLMs for making my beloved em-dashes and semicolons not cool.
—;—;—;;; —;—;—;—— —;———;;— ——;;;;;;
- Useless 'bot-skip' issue tag added for aura
Suggested policy
I would prefer a strict ban, such as that used by the Zig programming language. The primary developer of Zig is Andrew Kelley, who notably has contributed to LMMS in the past. Here's an interview where he explains some of his rationale for the strict ban: JetBrains — Zig 2026: No-AI Policy, $670K Foundation, Left GitHub & Why Zig Isn’t 1.0 - Andrew Kelley Explains
A slightly more lenient policy, such as that used by the Servo browser or the Rust Audio community, would also be acceptable to me.
I love LMMS. 💚
Discussion on Discord for additional context
https://discord.com/channels/203559236729438208/784594576223633478/1524114550469824714
I earnestly propose we adopt a "no-AI" policy for both technical and ethical reasons.
Readers may better recognize me by my YouTube handle, Cubician. I've been involved with LMMS in some form for over half my lifetime now: first as a pest on the mailing list, then a content creator, and currently, a developer. I don't make this suggestion lightly. LMMS is important to me. I do not think LLMs or other so-called "generative AI" are healthy for LMMS.
Technical reasons to avoid "AI"
LMMS has not had a stable release in six years at the time of writing. We're finally approaching a release of 1.3-alpha.2, only after a feature freeze full of dedicated bugfix efforts. LMMS has needed and continues to need thoughtful maintenance and refactoring more than anything else. I'm really proud of all the human effort that has gone into making LMMS more stable in the past few months.
Meanwhile, I continue to see people vibe-code features, including but not limited to: VST3 support, user interface overhauls, instrument plugins a human is already working on, and effect plugins we literally already have, the proompters of which then post demo videos, say "it's basically done", but never finish it. The LLM doesn't understand the output because it is incapable of reasoning (and can provide, at best, post-hoc rationale). The prompter doesn't understand the output because they didn't write it. It is the hardest kind of code to maintain: parasitic and verbose code nobody is accountable for, written with no intention whatsoever. This is fundamentally at odds with our needs as a project at this time, and fundamentally at odds with our future efforts at keeping LMMS healthy and maintainable for future developers.
But beyond personal anecdote, there is already real-world negative impact from using LLMs to generate code:
createPost()in increasingly erratic ways)And research showing that those who use LLMs to generate code suffer from "deskilling":
Ethical reasons to avoid "AI"
"Generative AI" as it is currently trained and offered is an inherently unethical tool. Stolen training data is fed into noxious datacenters, the construction of which is enabled by political corruption and in spite of civilian protest, driving up electricity, water, and consumer electronics prices for working people already struggling with the cost of living, the final product of which is used to collect data to enable citizen surveillance and war crimes. The math is morally neutral, but the material way it has been employed is morally bankrupt.
Environmental impact
Impact on hardware prices
Use in the war industry
Stolen training data
And the potential legal ramifications due to said stolen training data:
Miscellany
—;—;—;;; —;—;—;—— —;———;;— ——;;;;;;Suggested policy
I would prefer a strict ban, such as that used by the Zig programming language. The primary developer of Zig is Andrew Kelley, who notably has contributed to LMMS in the past. Here's an interview where he explains some of his rationale for the strict ban: JetBrains — Zig 2026: No-AI Policy, $670K Foundation, Left GitHub & Why Zig Isn’t 1.0 - Andrew Kelley Explains
A slightly more lenient policy, such as that used by the Servo browser or the Rust Audio community, would also be acceptable to me.
I love LMMS. 💚
Discussion on Discord for additional context
https://discord.com/channels/203559236729438208/784594576223633478/1524114550469824714
