An AI Policy for my music, writing and art

Why bother?

The truth is, I am afraid.

Well, that at least is what it came down to when I tried to track down why the desire to share my work faded more and more since generative AI became so prevalent on social media platforms.

I am sure you have seen it: Creatives needing to defend their work against accusations. Add to that the doubts if I really want to add my signal to the noise...

I have come across more and more cases recently, where genuine people had to defend themselves against the accusations that their work - or even literally them, their whole "persona" - were AI generated. The pinnacle of this is illustrators needing to screencap all of their drawing sessions to have proof that they actually did the work. Like, are you serious!? I will never cave in and do this for my music. My creative process is my own, and will remain so. Full stop. This is not me being secretive. This is me needing a safe space to be creative.

This scenario scares me more than the opposite case, where people defended work as their own that actually was wholly AI generated. That is just plain lying, and it is in my power to just ... not do that. But the former case is about public perception. And the current landscape where the "capabilities" of generative AI shift so frequently and are therefore misjudged makes the public perception of art subject to even more random waves of AI scepticism. And that scepticism is appropriate by the way!

But I, as a composer, and not least someone that sells their creativity on tap as a skill to collaborators, need to protect the perception of my art. Because I care for genuine art so so much, more than anything.

On writing online

Before I share with you my exact guidelines, please know that I would love for you to jump in to this discussion. I can't be alone with the fear that my original thoughts be attributed to some language model, robbing me of all my humanness, right? Right??

It's 18:31 on a Friday and I am listening to an ambulance drive by as I type this. I am not a content machine. I want to share what I care about.

But in the sea of slop on these platforms, I feel like I need to make sure that my writing is perceived that way. Or would it be enough to just ... write as humanly as possible? Is my fear actually a non-issue? Please do share your thoughts.

Johann's AI Policy

Effective September 1st, 2026. Do note that any and all of my previously published work follows these same guidelines. The licence text is dedicated to the public domain under licence CC0.

§ 1 Music

(1) For the intents and purposes of this personal AI Policy, the term "music" shall encapsulate both compositions (aka "sheet music") and musical recordings of compositions (aka "audio files").

(2) The following policy shall be followed:

1. No music will be generated by AI.

2. AI will not be used to "assist" during composition.

3. No parts of existing AI generated material will be sampled or adapted in any way to create new musical works.

4. Tools for sound design, mixing and mastering that claim to use machine learning can be used, as long as they offer at least some deterministic control over important parameters.^[1]

5. Tools for STEM-Separation can be used.

§ 2 Writing

(1) For the intents and purposes of this personal AI Policy, "writing" or "written work" shall be defined as (but not limited to) Essays, Blog or Social Media Posts, Articles, "Tweets", Scripts, Poems, Song Lyrics, Fiction, Non-Fiction etc. and extend to all forms of presentation, both in physical and digital media.

(2) The following policy shall apply:

1. No portion of any written work will be generated by a Large Language Model. Similarly, AI will not be used to draft, outline or re-phrase written work.

2. Grammar, spell checking and dictionary/translation tools can be used.^[2]

§ 3 Images and Videos

1. No still images or videos will be generated by AI.

2. Photo and Video editing tools that may use machine learning to enhance the workflow (i.e. transcriptions) or editing capabilities (i.e. background removal) can be used.^[3]

§ 4 Combined Works

1. Some works may combine different kinds of media. The above policy shall apply for all parts of those works respectively (e.g. the written script for a recorded podcast episode, the background music of a video, the images on a website, etc.).

2. When collaborating with other companies or people, I can only be held accountable for AI-use in my own contributions to the project. A more loose AI Policy from a party I work with must not water down my own, while a more strict one is allowed to make me match their standards for the project.

§ 5 Labeling

(1) Appropriate labeling of works as "human made" can be employed in a way that is idiomatic for the end product. The absence of such a label does not indicate any usage of AI beyond what this policy allows.(4) Usage of AI beyond this policy must be labeled and/or communicated.(3) As of August 2026, the brainmade.org mark is to be preferred.

§ 6 Updates

(1) This policy shall always be kept up to date and in line with my workflows and personal stance on the topic. Prior versions shall be accessible in archived form.

Footnotes

[1] I specifically use the word "claim" here, as I get the feeling that some tools jumped on the AI-Hypetrain without actually employing neural networks or machine learning at all, but actually still achieve their functionality with deterministic algorithms under the hood.

[2] Not least because their widespread integration into all kinds of software with constant updates makes it impractical to research their specific inner workings on a regular basis.

[3] Not least because their widespread integration into all kinds of software with constant updates makes it impractical to research their specific inner workings on a regular basis.