Arkancia is the first AI-native talent label. Our artists are fictional personas created and owned by Arkancia, produced with AI-assisted tools alongside human songwriters, producers, vocal engineers, and visual designers. This page explains, in plain language, how we use AI, what we refuse to do, and how we comply with the EU AI Act.

1. AI-assisted, human-directed

Every Arkancia release is directed by humans. AI tools are used for melody / arrangement scaffolding, vocal timbre synthesis (on original, non-identifiable voices only), visual concepting, and content production pipelines. Humans write the stories, approve every output, and own every rights package.

2. No voice cloning of identifiable persons

We do not clone the voice of any identifiable real person — living or dead — under any circumstance. No impersonation, no sound-alike product positioning, no ambiguity. Every vocal timbre used in an Arkancia release is an original synthesised voice tied to a fictional character Arkancia owns.

3. Disclosure per EU AI Act (2024/1689)

We disclose AI involvement on every release metadata tag, every artist page, and in the ISRC-linked release notes. This exceeds the minimum transparency requirements of Article 50 of the EU AI Act on content generated or significantly modified by AI.

4. Rights & clearance

All masters, publishing rights, and artist likeness are owned outright by Arkancia. Human collaborators are paid under standard work-for-hire or publishing splits, with transparent public credit. Sync licensing is therefore single-point: one signature, full clearance.

5. Training data provenance

For commercial output we only use models and assets with documented, licensable provenance — either licensed commercial models, first-party training data, or public-domain sources we can evidence. We don't deploy outputs we can't defend.

6. Questions?

Journalists, supervisors, and artists are welcome to ask for specifics on any individual release. Get in touch.