MyBunny AI — agentic threat model
MyBunny AI is a low-autonomy conversational companion platform with high non-determinism due to its unfiltered, creative nature. Its primary security risks center around user privacy, data exposure of sensitive NSFW interactions, and the lack of input/output guardrails.
OWASP AIVSS score rationale
| Autonomy of Action | 0.20 | |
| Goal-Driven Planning | 0.10 | |
| Self-Modification | 0.20 | |
| Dynamic Tool Use | 0.00 | |
| Persistent Memory | 0.40 | |
| Contextual Awareness | 0.30 | |
| Dynamic Identity | 0.10 | |
| Multi-Agent Interactions | 0.00 | |
| Non-Determinism | 0.80 | |
| Opacity & Reflexivity | 0.50 |
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 — likely utilizes open-source foundation models optimized for creative writing. The explicit lack of filters makes the model highly susceptible to adversarial jailbreaks, prompt injection, and generating extreme or misaligned outputs.
Not certain from the listing — the platform must store highly sensitive, potentially NSFW user chat histories and customization profiles. This creates a high-value target for data exfiltration, credential theft, or database leaks if proper encryption is not implemented.
Not certain from the listing — orchestration appears limited to basic chat memory management and system prompt customization. Risks include session state pollution and prompt injection that could permanently corrupt a companion's persona.
Not certain from the listing — being open-source and freemium, deployment could range from local hosting to multi-tenant cloud environments. Risks include insecure default configurations, lack of tenant isolation, and potential host compromise if self-hosted without sandboxing.
Not certain from the listing — the platform's focus on 'no unnecessary filters or restrictions' strongly implies a lack of real-time safety guardrails, input filtering, or output monitoring, leading to high exposure to toxic or harmful content generation.
Not certain from the listing — handling highly personal and NSFW companion data requires strict privacy compliance (e.g., GDPR, CCPA) and robust access controls, which are not detailed and may be lacking in self-hosted or early-stage deployments.
The platform operates as a standalone companion chat interface with no integration into a multi-agent ecosystem or external marketplaces mentioned, minimizing cascading ecosystem risks.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).
These scores are auto-generated from public information (the agent's own listing, docs, and repository) using the canonical OWASP AIVSS formula and the MAESTRO framework — an estimate for guidance, not a penetration test, audit, or certification. See the scoring methodology. Are you the vendor? Factual corrections are free.