Glm Image — agentic threat model
GLM Image is primarily a text-to-image generation model accessed via APIs rather than an autonomous agent, presenting low agentic risk. Its primary security concerns center on model-level vulnerabilities (adversarial prompts, output alignment) and API security rather than autonomous execution threats.
OWASP AIVSS score rationale
| Autonomy of Action | 0.10 | |
| Goal-Driven Planning | 0.00 | |
| Self-Modification | 0.00 | |
| Dynamic Tool Use | 0.00 | |
| Persistent Memory | 0.00 | |
| Contextual Awareness | 0.20 | |
| Dynamic Identity | 0.00 | |
| Multi-Agent Interactions | 0.00 | |
| Non-Determinism | 0.50 | |
| 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.
Utilizes a hybrid 9B autoregressive core and a 7B DiT diffusion decoder. Primary threats include adversarial prompt injection to bypass safety filters, model extraction/stealing via API querying, and generation of misaligned or harmful outputs.
Not certain from the listing — training data details are proprietary. Threats include potential copyright infringement in training data, data poisoning during pre-training, and lack of transparency in data lineage.
Not certain from the listing — GLM Image is primarily a model/API rather than an active agent framework. If integrated into orchestrators, threats include insecure tool integration and prompt injection via orchestrating frameworks.
Not certain from the listing — hosted as a cloud API with Python/Java SDKs. Threats include API key exposure, unauthorized API access, and potential denial of service on the generation endpoints.
Not certain from the listing — no built-in guardrails or observability tools are mentioned. Threats include a lack of input/output filtering for NSFW or policy-violating generations.
Not certain from the listing — compliance certifications (like SOC2 or GDPR) are not specified. Threats include compliance violations if used to generate copyrighted or deepfake material without consent.
Not certain from the listing — does not natively interact in a multi-agent ecosystem. Threats are limited to downstream applications consuming its API without proper validation.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).