"Agentic AI and Cyberattacks"

government

US Congress·congress.gov/crs-product/IF13151

Credibility Rating

5/5

Gold(5)

Gold standard. Rigorous peer review, high editorial standards, and strong institutional reputation.

Rating inherited from publication venue: US Congress

An official U.S. Congressional Research Service brief providing policymakers with a concise overview of agentic AI in military cyber operations; useful as a snapshot of the U.S. government's awareness of the governance gap surrounding autonomous AI systems.

Metadata

Importance: 62/100policy briefanalysis

Summary

This Congressional Research Service report examines agentic AI—autonomous systems operating with minimal human oversight—and its implications for offensive and defensive cyber operations. It surveys U.S. Department of Defense efforts (DARPA, NSA) to develop and test agentic AI capabilities, while highlighting the absence of formal government policy specifically governing these systems. The report underscores a growing governance gap as deployment outpaces regulatory frameworks.

Key Points

•Defines agentic AI as autonomous systems capable of pursuing goals with limited human supervision, distinguishing them from prior narrow AI tools.
•Outlines potential military cyber applications including automated vulnerability discovery, rapid offensive operations, and accelerated defensive response.
•Surveys DoD components—DARPA, NSA, Cyber Command—actively developing and testing agentic AI cyber capabilities.
•Identifies a significant policy vacuum: no U.S. government regulations or directives specifically address agentic AI in cyber contexts as of the report's publication.
•Raises implicit concerns about accountability, escalation risk, and the speed of autonomous cyber action outpacing human decision-making.

Cited by 1 page

Page	Type	Quality
AI Alignment	Approach	91.0

Cached Content Preview

HTTP 200Fetched Mar 15, 202616 KB

[skip to main content](https://www.congress.gov/crs-product/IF13151#content)

[Citation](https://www.congress.gov/crs-product/IF13151#citation)[Subscribe](https://www.congress.gov/get-alerts)[Share/Save](https://www.congress.gov/crs-product/IF13151#share)[Site Feedback](https://www.congress.gov/crs-product/IF13151#survey)

# Agentic Artificial Intelligence and Cyberattacks

## CRS PRODUCT (LIBRARY OF CONGRESS)

[Hide Overview](https://www.congress.gov/crs-product/IF13151#)

| CRS Product Type: | In Focus |
| CRS Product Number: | IF13151 |
| Referenced Legislation: | [P.L.114-113](https://www.congress.gov/bill/114th-congress/house-bill/2029/text); [P.L.119-37](https://www.congress.gov/bill/119th-congress/house-bill/5371/text); [P.L.119-60](https://www.congress.gov/bill/119th-congress/senate-bill/1071/text) |
| Topics: | Defense & Intelligence; Science & Technology |
| Publication Date: | 02/03/2026 |
| Authors: | Sayler, Kelley M.; Theohary, Catherine A. |

[Download PDF (216KB)](https://www.congress.gov/crs_external_products/IF/PDF/IF13151/IF13151.2.pdf)
\| [PDF Version History](https://www.congress.gov/crs-product/IF13151#)

- [February 3, 2026 (IF13151 - Version: 2)](https://www.congress.gov/crs_external_products/IF/PDF/IF13151/IF13151.2.pdf)
- [January 14, 2026 (IF13151 - Version: 1)](https://www.congress.gov/crs_external_products/IF/PDF/IF13151/IF13151.1.pdf)

[Listen](https://app-na.readspeaker.com/cgi-bin/rsent?customerid=6927&lang=en_us&readid=dvContainer&url=https%3A%2F%2Fwww.congress.gov/crs-product/IF13151 "Listen to this page using ReadSpeaker webReader")

Agentic Artificial Intelligence and Cyberattacks

Updated February 3, 2026
(IF13151)


Introduction

_Agentic_ means autonomous, or independent. Agentic artificial intelligence (AI) capabilities are of increasing interest to the U.S. military and to Congress. According to an [IBM definition](https://www.ibm.com/think/topics/agentic-ai), "Agentic AI is an artificial intelligence system that can accomplish a specific goal with limited supervision. It consists of AI agents—machine learning models that mimic human decision-making to solve problems in real time.... Unlike traditional AI models, which operate within predefined constraints and require human intervention, agentic AI exhibits autonomy, goal-driven behavior and adaptability." For an explanation of AI- and machine learning-related terms, see CRS Infographic IG10077, [_Artificial Intelligence (AI) Taxonomy_](https://www.congress.gov/crs-product/IG10077), by Laurie Harris and Nora Wells. According to the Department of Defense (DOD)—which is "using a secondary Department of War designation" under [Executive Order](https://www.federalregister.gov/documents/2025/09/10/2025-17508/restoring-the-united-states-department-of-war) 14347 dated September 5, 2025—Cybersecurity and Information Systems Information Analysis Center, there are "no known official government guidance or policies yet specifically on agentic AI."

Agentic AI and Def

... (truncated, 16 KB total)

Resource ID: a6df5717790c33c9 | Stable ID: OGZiMzkxNW