The Declarant has assessed the Orbit Voice AI System against the risk classification criteria of Regulation (EU) 2024/1689 (the "EU AI Act") and has determined as follows:
| Classification Category | Applicable? | Basis |
|---|---|---|
| Unacceptable Risk (Article 5 — prohibited) | No | System does not engage in social scoring, subliminal manipulation, real-time biometric surveillance in public spaces, or other prohibited practices |
| High Risk (Annex III) | No | System is not used in critical infrastructure, education, employment decisions, essential services, law enforcement, migration control, or administration of justice |
| Limited Risk — Article 52 Transparency Obligations | YES | System interacts with natural persons via voice and must disclose its AI nature. Article 52(1) applies: persons must be informed they are interacting with an AI system. |
| Minimal Risk (General Purpose AI) | Partial | Underlying LLM (Anthropic Claude) is a general-purpose AI model subject to GPAI transparency obligations; BizBot as deployer bears Article 52 obligations. |
Conclusion: The Orbit Voice AI System is classified as a Limited Risk AI System subject exclusively to the transparency obligations of Article 52 of the EU AI Act. It is NOT a high-risk AI system and is NOT subject to the conformity assessment procedures of Articles 43–49.
The Declarant hereby declares that the following Article 52 transparency obligations have been implemented:
| Obligation | Status | Implementation Evidence |
|---|---|---|
| AI Disclosure to End Users (Art. 52(1)) Persons interacting with AI must be informed of the AI nature of the system before first interaction or at the start of interaction |
✓ IMPLEMENTED | Voice disclosure script active in api/retell-call.js. Orbit identifies itself as an AI assistant at the start of every call. Disclosure language visible at /voice.html. |
| AI System Identification on Request (Art. 52(1)) Deployer must be able to identify the AI system upon competent authority request |
✓ IMPLEMENTED | System registry available at api/ai-system-registry.js. Logs AI system name, version, provider (Retell AI + Anthropic Claude), and deployment context. |
| Incident Reporting Procedure (Art. 73 / Art. 52 context) Mechanism to detect, log, and report serious incidents |
✓ IMPLEMENTED | Incident reporting endpoint active at api/ai-incident-report.js. Logs incidents to Google Sheets audit log with severity classification, timestamp, and description. |
| Human Oversight Mechanism Users must be able to opt out of AI interaction and reach a human |
✓ IMPLEMENTED | Live call transfer to human staff available at any point during AI interaction. End users can request human handoff verbally. Opt-out is immediate and unconditional. |
| No Deceptive AI Persona AI system must not impersonate a human when directly and sincerely asked |
✓ IMPLEMENTED | Orbit is programmed to confirm its AI nature if directly asked by a caller. System prompt prohibits claiming to be human. Tested and verified. |
The Orbit Voice AI System processes personal data (voice recordings, caller identity, call content) in connection with its operation. The Declarant confirms:
This declaration has been prepared with reference to:
This Declaration of Conformity reflects the state of the Orbit Voice AI System as of the date of signature. BizBot Technology LLC does not warrant that the AI system will perform without error in all circumstances. The Declarant will update this declaration upon any material change to the AI system's functionality, risk profile, or the regulatory requirements applicable to it.
This document does not constitute legal advice and is not a substitute for competent EU legal counsel where the EU AI Act applies to specific deployments.
The undersigned hereby declares, on behalf of BizBot Technology LLC, that the Orbit Voice AI System described in this document complies with the applicable transparency obligations of Article 52 of Regulation (EU) 2024/1689 as of the date of signature.