Neuro-Agentic Control: A Deep Learning-based LLM-Powered Agentic AI Framework for Controlling Security Controls
Saroj Gopali, Bipin Chhetri, Deepika Giri, Sima Siami-Namini, Akbar Siami Namin
Why It Matters
What makes this one worth your time
This research is relevant for AI engineers and researchers focused on improving the security and reliability of industrial IoT systems using advanced AI techniques.
A novel framework using LLMs and foundation models to enhance security in industrial IoT environments.
Summary
The paper presents a novel neuro-agentic control framework that integrates a large language model-based planner with a time-series foundation model to enhance autonomous defense in industrial IoT environments. It introduces a 'Counterfactual Physics Injection' mechanism to simulate and evaluate LLM-proposed interventions, preventing unsafe actions. The framework is tested on an industrial dataset, showing improved performance over baseline models.
Key contributions
- Introduction of a neuro-agentic control framework for industrial IoT security.
- Development of a 'Counterfactual Physics Injection' mechanism to evaluate LLM interventions.
- Demonstrated improved performance over LSTM and TCN baselines in preventing security breaches.
Notable insights
- The use of 'Counterfactual Physics Injection' to simulate interventions in a latent space is a clever method to prevent unsafe actions.
- Combining LLMs with time-series models for security control is an innovative approach to address the limitations of traditional rule-based systems.
Possible limitations
- Not stated in the abstract
Abstract
arXiv:2607.09076v1 Announce Type: new Abstract: Cyberattacks on operational technology are increasingly causing costly downtime and physical damage, exposing the limitations of traditional rule-based monitoring in industrial IoT environments. While Large Language Models (LLMs) have strong semantic reasoning abilities to assist in decision support, their hallucinatory nature presents unacceptable safety liabilities for closed-loop control. This paper introduces a neuro-agentic control framework, a novel architecture that couples an LLM-based planner (i.e., such as Gemini 2.5 Flash-Lite) with a pre-trained Time-Series Foundation Model (TimesFM), to achieve physics-grounded autonomous defense. The paper introduces a ``Counterfactual Physics Injection'' mechanism that simulates the impact of LLM-proposed interventions within the numerical latent space of the foundation model before actuation, while allowing the system to reject hallucinatory or unsafe actions. Evaluated on an industrial dataset (e.g., the Secure Water Treatment (SWaT)) in the context of stochastic attack scenarios, the framework exhibited better performance compared to LSTM and TCN baselines. The Neuro-Agentic Loop prevented five breaches (33.3%) below the threshold versus LSTM (26.7%) and TCN (13.3%), with zero physically invalid (hallucinated) actions executed. These results demonstrate the efficacy of using foundation models as deterministic ``Sentinels'' to safeguard agentic AI in critical infrastructure.