Leonardo AI — agentic threat model
Leonardo AI exhibits low agentic risk due to its limited autonomy, planning, and tool-use capabilities, operating primarily as a human-in-the-loop generative content tool. Its primary security risks reside in model safety (bypassing content filters) and the protection of proprietary model weights and user-uploaded training data.
OWASP AIVSS score rationale
| Autonomy of Action | 0.10 | |
| Goal-Driven Planning | 0.10 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.10 | |
| Persistent Memory | 0.30 | |
| Contextual Awareness | 0.20 | |
| Dynamic Identity | 0.00 | |
| Multi-Agent Interactions | 0.00 | |
| Non-Determinism | 0.80 | |
| Opacity & Reflexivity | 0.70 |
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.
Uses proprietary and fine-tuned diffusion models (and potentially LLMs for prompt expansion). Threats include adversarial prompt injection to bypass safety filters, model stealing/extraction of proprietary weights, and output misalignment such as generating unauthorized or harmful content.
Not certain from the listing — Data operations likely involve training and fine-tuning pipelines on user-uploaded images and proprietary datasets. Threats include training data poisoning, copyright infringement risks, and unauthorized access to user-uploaded assets.
Not certain from the listing — The orchestration layer likely manages prompt processing and model inference pipelines rather than complex agentic planning. Threats include insecure handling of user prompts and potential injection attacks in prompt-enhancement features.
Not certain from the listing — Infrastructure likely relies on high-performance GPU cloud hosting. Threats include container escape, unauthorized access to model weights, and API abuse leading to resource exhaustion or denial of service.
Not certain from the listing — Observability likely focuses on generation quality, latency, and content moderation filters. Threats include bypass of safety guardrails (e.g., generating deepfakes or CSAM) due to blind spots in automated moderation.
Not certain from the listing — Compliance posture regarding data privacy (GDPR/CCPA) and intellectual property rights for AI-generated art is unclear. Threats include regulatory non-compliance and lack of robust user access controls.
Not certain from the listing — The platform operates primarily as a standalone SaaS tool rather than a multi-agent ecosystem. Threats are limited to API integrations and potential marketplace abuse if custom models are shared.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).