AI Systems Register
This register lists all AI systems Orbirank uses to generate content, make decisions, or present recommendations to users. Published in accordance with the EU AI Act (Article 52, GPAI transparency obligations), effective August 2, 2026. Updated whenever a new model version is deployed to production.
Every piece of AI-generated content shown to users carries one or more of the following labels:
| Label | Where shown | Meaning |
|---|---|---|
| AI-Generated badge | Content draft cards, brief previews | The text was produced by an LLM, not a human author. |
| Review before publishing banner | Every content draft review screen | Cannot be dismissed. Reminds users to verify accuracy before publishing. |
| Confidence badge (High / Medium / Low) | Content variant cards | Heuristic confidence based on source citations found and verified. Low confidence blocks one-click apply. |
HTML comment: <!-- ai-assisted-content: true --> |
CMS-injected HTML output | Machine-readable label that survives CMS publish, for downstream tool transparency. |
The disclosure label is shown to all users, regardless of location. EU/EEA users additionally see a one-time acknowledgment prompt on first content generation explaining AI involvement in Orbirank's outputs.
Orbirank's AI systems are classified as Limited Risk (Article 52 GPAI) and Low Risk (general-purpose text classification). None of Orbirank's AI systems fall under the High Risk categories defined in Annex III of the EU AI Act (2024/1689). Specifically:
- No use in critical infrastructure, education, employment, or access to essential services.
- No AI system makes autonomous high-consequence decisions without user review.
- No use of biometric identification, emotion recognition, or social scoring.
- Content generation is advisory only — users control publication.
| Data input | Purpose | Retained by LLM provider? |
|---|---|---|
| Your website's crawled page content | Generate content briefs + optimised copy | No — API calls; Anthropic's zero-data-retention policy for API |
| Target keywords (user-supplied) | Content and audit analysis | No |
| SEO audit findings (issue text) | Generate fix recommendations | No |
| Competitor page snippets (SERP) | Content gap analysis + brief generation | No |
Orbirank uses Anthropic's API under its commercial terms. Anthropic does not train on API inputs by default. See Anthropic Privacy Policy for full details.