Augmenting LLM-Based Agents for Improved Performance in Pentesting and Commissioning Operational Technology in Critical Infrastructure

Werth, Aaron, Hahn, Gary, Borges Hink, Raymond et al. · 2026

Read Augmenting LLM-Based Agents for Improved Performance in Pentesting and Commissioning Operational Technology in Critical Infrastructure by Werth, Aaron, Hahn, Gary, Borges Hink, Raymond et al. (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/doe-osti-3367033-augmenting-llm-based-agents-for-improved.txt.

Description

Oak Ridge National Laboratory report ORNL/TM-2026/4611, by Aaron Werth, Gary Hahn, Raymond Borges Hink, and six co-authors, tests whether language-model agents can help attack and commission industrial control systems. The same agent that penetration-tests a plant network might also help operators configure it. Critical infrastructure needs both capabilities tested safely before adversaries improvise their own.

The June 2026 study augments LLM-based agents with tools and context for operational technology work, then measures their performance on pentesting and commissioning tasks. Findings speak to where agents already help, where they stumble on protocol specifics, and what guardrails matter when the target is live infrastructure rather than a capture-the-flag game. Utilities and labs weighing AI assistants for control systems get evidence instead of vendor claims.

Catalog record

CREATORWerth, Aaron, Hahn, Gary, Borges Hink, Raymond et al.
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52 KB
LICENSEPublic Domain
RIGHTSPUBLIC DOMAIN
SOURCEDOE OSTI

Source: https://www.osti.gov/biblio/3367033

Browse the full library · About the Tower of Babel