AI Disclosure

1. Why This Page Exists

Article 50 of the EU Artificial Intelligence Act requires providers and deployers of general-purpose AI systems to mark AI-generated output in a machine-readable format and accompany it with a human-readable disclosure when it is shown to people in the European Union. MusicOps serves independent Latin labels whose audiences include EU listeners, so we apply these obligations to every AI- generated surface, regardless of the viewer's location.

2. Where AI Is Used in MusicOps

• Music Brain (track analysis). Uploaded audio is analyzed by Google Gemini 2.5 Flash (musical features: BPM, key, mood, energy, hooks, sections, instruments), OpenAI Whisper (lyric transcription), and Anthropic Claude Sonnet 4.6 (strategic synthesis: theme, audience type, platform fit, content angles). All output is presented as suggestions, never as facts about the listener.
• AI Feed. Daily insights are produced by rule-based detectors over your release, content, audience, and pulse data, then formatted by Anthropic Claude Haiku 4.5 for readability. Each item shows its source engines.
• Weekly content pack. Each Monday, social post ideas, podcast angles, and newsletter subject lines are drafted by Anthropic Claude Haiku 4.5 for your active releases, optionally informed by Music Brain output.

3. How We Mark AI Output

Every AI-generated surface carries two markers:

• Machine-readable. API responses include an ai_provenance object listing the model(s) used, the generation timestamp, and whether a human has reviewed the output. DOM nodes that wrap AI output set data-ai-generated="true" with the model name. The document head includes <meta name="ai-disclosure">.
• Human-readable. A visible badge (subtle inline pill or full banner) appears next to AI output. Hovering or tapping the badge reveals the model name, generation date, and a reference to EU AI Act Article 50.

4. What We Do Not Do

• We do not deploy emotion-recognition or biometric categorization systems on users or artists.
• We do not generate synthetic likenesses of real people (deepfakes) of any kind. AI output is limited to text and structured data about music tracks and release strategy.
• We do not operate any system that meets the EU AI Act high-risk classification (Annex III). No registration under Article 51 is required for MusicOps as of this disclosure's effective date.

5. Opt-Out

AI features (Music Brain analysis, AI Feed generation, weekly content pack) are opt-in per organization. Settings > AI & Data lists every model used, its purpose, and a control to turn the feature off for your org.

6. Updates

We will revise this page when we add or remove AI models, when EU AI Office guidance changes the meaning of Article 50, or when MusicOps adds new AI-generated surfaces. The effective date above always reflects the most recent substantive change.

7. Contact

Questions about AI use in MusicOps: [email protected].