I'm debating another round of 50 lyrics in 3 months for a GRTX application. All the writing is original and authored by me without any assistance. This matters because at some point it will all be fed into the AI in some way to experiment and create music. A good portion of it is performed by me first (sometimes with midi or vocal imitations of instrumentals), creating that 100% human first fixation before using it as audio input for the AI.
Here's the main issue, extremists have hijacked the conversation about AI and labeled all use as equal to one-prompt Generative AI. The Flat Earth Cultists are demanding purity, to the point that original human authored work is being flagged and erased. A poem published in 2012 performed with an added hook and used as input is treated as though it erases all the human effort and rights. That is where we are now, and we need to start pushing back. Many that claim AI is erasing human works have become the worst offenders for attempting to erase human authored expression. The arts are not under attack by AI, it is under attack by the extremists on both sides.
Now I do have a dilemma in regards to the new labeling systems. Originally everything I wanted to do was going to be assistive-AI, but now I'm creating some 100% human attributed works utilizing DAW tools intended to mimic AI. I want those works to even trip the detectors. Blurring that line is part of my expression and it isn't the lowercase ai label that bothers me, it's having to essentially label my non-AI work as purely human. It makes me want to sample a 2 second AI instrument to mix in just so I can call it an AI song.
What will happen is that the 100% human prototype / demo is going to be either copyrighted or publicly released before an AI version goes public. That PA/SR copyright has some added value because it gives recording rights. When a first authorized recording of a song goes public it triggers Section 115 for compulsory license. It lets my song get covered and collect royalties. It also turns that AI version into a derivative version which operates outside of cover licenses. Copying version 2 becomes a firm violation of the copyright on version 1, and because version 2 doesn't trigger a second compulsory license they can't use the AI to erase my rights on version 1. If I don't secure copyright on version 1 first they basically attempt to strip my authorship from both versions.
GRTX lets me separate the lyrical copyright from the musical copyright. If I do GRUW for group recordings of my version 1's before public release, I disclaim the lyrics as previously published and cover the musical composition and recording separately from the lyrics as new, unpublished works. That way if someone tries to use just the lyrics in a completely new composition I can focus on just the literary element, which for me has always been 100% human authored. The courts would hopefully limit the case to only the literary component so that the later use of AI isn't argued as erasing my rights.