SingGuard-NSFA: Extensible Guardrails for Agentic AI via Generative Reasoning and Real-Time Classification
SingGuard Team
Why It Matters
What makes this one worth your time
As AI systems become more autonomous, ensuring their security against operational threats is crucial for safe deployment in real-world applications.
SingGuard-NSFA enhances AI security with a novel guardrail framework achieving superior threat detection.
Summary
The paper introduces SingGuard-NSFA, a framework designed to secure agentic AI systems against operational threats using a taxonomy of 185 risk variants. It combines generative reasoning and real-time classification to detect threats, achieving high F1 scores on benchmarks and outperforming existing guardrails.
Key contributions
- Development of a comprehensive taxonomy organizing 185 risk variants.
- Creation of a benchmark suite with over 93K samples for evaluating AI security.
- Introduction of a dual-mode threat detection approach combining generative reasoning and real-time classification.
Notable insights
- The use of a dual-mode approach combining generative reasoning with discriminative classification for threat detection is innovative.
- The framework's extensibility allows it to adapt to new threats beyond its original design.
Possible limitations
- Not stated in the abstract
Abstract
arXiv:2607.13081v1 Announce Type: cross Abstract: We present nsfaguard, a guardrail framework for securing agentic AI systems against operational threats, such as prompt injection, sensitive information extraction, malicious code requests, dangerous tool misuse, and resource exhaustion. We first introduce the NSFA taxonomy, which organizes 185 risk variants into a CIA-triad-grounded hierarchy and is cross-validated against three well-established OWASP guidelines. Based on this taxonomy, we construct a benchmark suite spanning 133 languages, comprising over 93K purpose-built samples targeting both user queries and agent responses, along with 3,435 cross-source samples adapted from five public agent-security datasets. To detect these operational threats in practice, we develop a dual-mode approach combining SFT-based generative reasoning for interpretable offline auditing with discriminative classification heads on the frozen backbone, enabling real-time detection at approximately 50,ms. We release four models with 0.8B, 2B, 4B, and 9B parameters, all achieving $\geq$94% F1 on purpose-built benchmarks and surpassing the strongest competing guardrails by 6 to 12 absolute points. On cross-source evaluation, the 9B model attains 91.29% F1 with a more balanced precision--recall trade-off. Moreover, ablation experiments show that classification heads can equip a guardrail with risk detection capabilities beyond its original scope and achieve state-of-the-art performance. These results demonstrate the extensibility of the approach and its generality as a plug-in enhancement.