Skip to content
AI Music Video Generator

AI-generated content

Last updated: April 28, 2026

The app uses open-source AI models that run locally on your machine. They generate lyrics, images for the backgrounds you see, and the conversational responses in the Brand voice coach. They also analyze your lyrics to drive emphasis-aware effects.

These models are powerful and flexible, but unlike some commercial cloud services, they are not safety-trained. Output is not always predictable. This page explains what we do to keep results appropriate and where your responsibility starts.

What the app does to help

We’ve built guardrails into every generation pipeline:

  • Negative prompts and content filters are applied to every image generation request to discourage nudity, gore, hate symbols, graphic violence, and other clearly inappropriate content.
  • Prompt templates are scoped to music-video aesthetics — cinematic scenes, performers, environments, abstract visuals — rather than open-ended generation.
  • The Brand voice coach is scope-locked to brand-setup topics. It refuses off-topic requests, jailbreak attempts, and prompts that try to redirect it to other tasks.
  • Lyric generation is brand-scoped to your configured voice and style settings, which gives the model a consistent frame to work in.

What the app cannot guarantee

Despite these guardrails, occasional unexpected, low-quality, or off-tone output is possible. Open-source models are trained on broad internet-scale datasets and can produce surprises. Common categories of “off” output include:

  • An image with the wrong number of fingers, distorted faces, or other anatomical artifacts
  • A scene that doesn’t match the lyric’s mood
  • A lyric phrase that’s awkward, repetitive, or off-brand
  • An emphasis detection result that flags the wrong section
  • Brand-coach replies that occasionally drift or repeat

None of this means the app is broken — it means the underlying models are imperfect. Regenerate the affected output, adjust your brand prompt or video style, or edit the result downstream.

Your responsibility

Because the app runs locally and we have no visibility into what you generate or publish:

  • You are responsible for everything you publish under your name. Review every render — backgrounds, lyrics, and the final video — before you post it to YouTube, TikTok, Instagram, or any other platform.
  • You are responsible for compliance with the platforms you upload to. Each platform has its own content policies, and acceptance or removal of your content is between you and them.
  • You are responsible for any rights or clearances for the audio you supply to the app and for the personas, names, and likenesses referenced in your brand voice.

We don’t moderate, store, log, or review your output. We can’t be responsible for content the local models produce in your installation.

When something is genuinely broken

If you see content slipping through the guardrails consistently — for example, the negative-prompt filter not catching a clearly inappropriate image type, or the Brand voice coach engaging with off-topic prompts it should refuse — that’s a bug we want to fix. Please file an issue through the marketplace or seller you bought the app from. They’ll route it to us.

Isolated misfires from a model that’s working within its safety prompt are not bugs — they’re a known characteristic of open-source generation.