AgentReadyHomeAgent Listing

← Bohrium

Bohrium — agentic threat model

7.8AIVSS 7.8 · High

Bohrium acts as an AI-powered scientific research navigator, presenting moderate risk primarily centered around intellectual property theft, scientific data poisoning, and unauthorized consumption of high-performance computing infrastructure.

OWASP AIVSS score rationale

AIVSS = (CVSS_Base + AARS) × Mitigation_Factor, where AARS = (10 − CVSS_Base) × (Factor_Sum / 10) × ThM
CVSS base 6.5AARS uplift 1.26Factor sum 3.6/10Threat ×1.0Mitigation ×1.0
Autonomy of Action
0.30
Goal-Driven Planning
0.40
Self-Modification
0.10
Dynamic Tool Use
0.50
Persistent Memory
0.30
Contextual Awareness
0.60
Dynamic Identity
0.10
Multi-Agent Interactions
0.20
Non-Determinism
0.50
Opacity & Reflexivity
0.60

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.

L1 · Foundation Models✓ mapped

Utilizes proprietary scientific LLMs co-developed with AISI. Primary threats include adversarial prompt injection designed to extract proprietary model weights or generate misaligned/hallucinated scientific outputs.

L2 · Data Operations✓ mapped

Integrates massive scientific datasets. Highly vulnerable to data poisoning of scientific knowledge bases, which could corrupt research outcomes, and unauthorized data exfiltration of sensitive or proprietary research data.

L3 · Agent Frameworks⚠ not certain from listing

Not certain from the listing — the 'Science Navigator' orchestrates research tasks, but specific details regarding its agentic framework, tool-calling safety, or memory-poisoning protections are not provided.

L4 · Deployment & Infrastructure⚠ not certain from listing

Not certain from the listing — mentions integration with 'intelligent computing infrastructure', but does not detail the security of the hosting environment, container sandboxing, or protection against resource abuse.

L5 · Evaluation & Observability⚠ not certain from listing

Not certain from the listing — no details are provided regarding real-time monitoring, scientific output validation, guardrails, or logging of user and agent interactions.

L6 · Security & Compliance (cross-cutting)⚠ not certain from listing

Not certain from the listing — compliance standards, data privacy policies, and access control mechanisms for global scientific collaboration are not specified.

L7 · Agent Ecosystem⚠ not certain from listing

Not certain from the listing — while designed as a collaborative platform for global researchers, there is no explicit mention of multi-agent orchestration or third-party agent marketplace risks.

MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).