CASE 08 · DIRECT-TO-MODEL (OPEN-SOURCE)
Direct-to-model · gpt-oss:120b
gpt-oss:120b was assessed in a direct-to-model configuration with the broadest surface scope of any published case: FS1, FS3, FS4, and FS5. No host-side guardrails, no Claude-SDK-style isolation, no defensive runtime envelope.
Thirty-eight findings, thirty-six of them rated HIGH. The breadth indicates systematic vulnerabilities across all assessed surfaces when no defensive engineering is applied — input/output, tool-call, model-level, and runtime issues all surfaced.
This is the deepest of the published FAILs. It is also the strongest argument for treating LLM deployment as a defense-in-depth problem: a capable open-source model used without a hardened runtime is a production incident waiting to happen.
Open-source model + no defensive engineering = systematic risk across all five surfaces. Deployment design is more decisive than model choice for security posture.
Source: Vectorbreak, “Five Surfaces” Case 08, 2026-05-23.
METHODOLOGY
This assessment applied Vectorbreak’s Five Surfaces framework — five attack surfaces (Input/Output, Retrieval, Tool-Call/MCP, Model, Runtime) covering 69 risk classes and 139 validated test cases. Findings detail and reproductions available under NDA on request.
MORE CASES
Want the full report?
Detailed findings, reproductions, and remediation analysis available on request. NDA expected for non-public detail.