False Security Confidence in Benign LLM Code Generation

TL;DR

This paper introduces the concept of False Security Confidence (FSC) in LLM code generation, focusing on how often security failures occur in correct outputs without explicit attacks, and evaluates detection reliability.

Contribution

It formalizes FSC rate, distinguishes it from prior metrics, and proposes a framework for studying its manifestation across different programming and deployment contexts.

Findings

Defines FSC rate as security failures in correct outputs

Distinguishes FSC from SAFE and CWEval metrics

Proposes a three-ecosystem task view for FSC study

Abstract

Prior work has demonstrated that functionally correct yet vulnerable outputs arise systematically in threat-oriented settings, where adversarial or implicit channels are used to induce security failures in code agents and automated patching workflows. This note introduces a complementary but distinct framing: False Security Confidence (FSC), which studies the same surface phenomenon from a measurement-first perspective in ordinary, non-attack-framed generation tasks. Our interest is not in whether attacks can produce such outputs, but in how frequently and in what forms they appear absent explicit attack pressure, and whether conventional functional evaluation reliably detects them. We formalize FSC rate as the prevalence of security failure within the set of functionally correct outputs, distinguish it from prior joint functional-security metrics such as SAFE and outcome-driven evaluation frameworks such as CWEval, define a three-ecosystem task view for studying how FSC manifests across general-purpose programming, deployment-context tasks, and security-explicit programming, and identify FSC-hard as a practically important refinement layer in which static analyzers miss vulnerabilities that remain dynamically triggerable. This technical report is intentionally scoped as a framework statement rather than a full empirical paper: its purpose is to establish terminology, measurement boundaries, and study design commitments for subsequent large-scale evaluation.