Cal AI — agentic threat model
Cal AI presents a moderate-to-high risk profile due to its direct integration with enterprise calendars and its voice-based autonomous scheduling capabilities, which could be exploited for social engineering or unauthorized calendar manipulation.
OWASP AIVSS score rationale
| Autonomy of Action | 0.70 | |
| Goal-Driven Planning | 0.50 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.60 | |
| Persistent Memory | 0.30 | |
| Contextual Awareness | 0.60 | |
| Dynamic Identity | 0.20 | |
| Multi-Agent Interactions | 0.10 | |
| Non-Determinism | 0.50 | |
| Opacity & Reflexivity | 0.40 |
Scored with the canonical OWASP AIVSS formula (AIVSS calculator reference); agentic risk factors estimated from the agent’s described capabilities.
MAESTRO 7-layer threat model
Per-layer threats for this agent. Layers tagged “not certain from listing” are general, caveated commentary where the public description didn’t pin that layer.
Not certain from the listing — The specific LLM or voice synthesis models powering the voice agent are not disclosed, leaving potential vulnerabilities to adversarial voice cloning, prompt injection, or model reprogramming unverified.
Handles sensitive calendar data, availability, and user contact information. Risks include unauthorized data exfiltration of meeting details or calendar poisoning to manipulate availability.
Orchestrates voice calls and calendar tool execution. Vulnerabilities include insecure tool integration where malicious voice inputs could trick the agent into deleting or modifying critical calendar events.
Integrates with external calendar APIs (Google, Outlook) and telephony infrastructure. Risks involve API key exposure, insecure webhook endpoints, and potential lateral movement into connected enterprise calendar systems.
Not certain from the listing — There is no mention of real-time voice call monitoring, transcript logging, or guardrails to detect and prevent prompt injection during live phone calls.
Mentions 'Enterprise-level user management' which suggests role-based access controls (RBAC), but lacks explicit details on compliance standards (e.g., SOC2, HIPAA) or audit logging capabilities.
Not certain from the listing — While 'Cal atoms' are mentioned for integration, it is unclear if the agent interacts autonomously with other external AI agents or marketplaces, risking cascading trust failures.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).