



Network Working Group                                            Q. Yuan
Internet-Draft                                                    J. Mao
Intended status: Standards Track                                  B. Liu
Expires: 5 May 2026                                              N. Geng
                                                                X. Shang
                                                                  Q. Gao
                                                                   Z. Li
                                                     Huawei Technologies
                                                         1 November 2025


               Use cases of the AI Network Security Agent
               draft-yuan-rtgwg-security-agent-usecase-00

Abstract

   Core network devices like routers fulfill dual roles of data
   forwarding and security protection.  However, escalating threats
   (e.g., zero-day vulnerabilities, DDoS attacks) expose limitations of
   traditional security—relying on static ACLs, signature-based
   detection, and manual configuration—causing delayed responses, high
   false positives, and protection gaps.  This paper proposes AI Network
   Security Agents: intelligent software components leveraging machine
   learning, behavioral analysis, and real-time data fusion, with three
   core capabilities (adaptive learning, automation, distributed
   collaboration) to shift security from passive to intelligent.  Four
   key scenarios are outlined: dynamic defense against unknown threats
   via baselines and tracing; ACL optimization via intent parsing;
   configuration security via baseline checks and simulation; and
   collaborative defense via intelligence aggregation and linked
   responses.  AI Agents turn routers into active security
   orchestrators, enhancing threat protection and operational
   efficiency.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at https://datatracker.ietf.org/drafts/current/.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."



Yuan, et al.               Expires 5 May 2026                   [Page 1]

Internet-Draft           Security Agent Usecase            November 2025


   This Internet-Draft will expire on 5 May 2026.

Copyright Notice

   Copyright (c) 2025 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents (https://trustee.ietf.org/
   license-info) in effect on the date of publication of this document.
   Please review these documents carefully, as they describe your rights
   and restrictions with respect to this document.  Code Components
   extracted from this document must include Revised BSD License text as
   described in Section 4.e of the Trust Legal Provisions and are
   provided without warranty as described in the Revised BSD License.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
   2.  Requirements Language . . . . . . . . . . . . . . . . . . . .   3
   3.  Usage Scenarios of the AI Network Security Agent  . . . . . .   3
     3.1.  Dynamic Defense Against Zero-Day Vulnerabilities and
           Unknown Threats . . . . . . . . . . . . . . . . . . . . .   3
       3.1.1.  Behavioral Pattern Recognition  . . . . . . . . . . .   3
       3.1.2.  Attack Chain Tracing  . . . . . . . . . . . . . . . .   4
       3.1.3.  Zero-Day Vulnerability Prediction . . . . . . . . . .   4
     3.2.  Dynamic ACL Rule Optimization and Intelligent Policy
           Management  . . . . . . . . . . . . . . . . . . . . . . .   4
       3.2.1.  Policy Intent Translation . . . . . . . . . . . . . .   4
       3.2.2.  Traffic Behavior Learning . . . . . . . . . . . . . .   4
       3.2.3.  Policy Conflict Detection . . . . . . . . . . . . . .   4
     3.3.  Device Configuration Security . . . . . . . . . . . . . .   4
       3.3.1.  Configuration Baseline Verification . . . . . . . . .   5
       3.3.2.  Configuration Change Validation . . . . . . . . . . .   5
     3.4.  Threat Intelligence Sharing and Global Collaborative
           Defense . . . . . . . . . . . . . . . . . . . . . . . . .   5
       3.4.1.  Intelligence Aggregation  . . . . . . . . . . . . . .   5
       3.4.2.  Linked Attack Response  . . . . . . . . . . . . . . .   5
       3.4.3.  Security Posture Prediction . . . . . . . . . . . . .   5
   4.  Conclusion  . . . . . . . . . . . . . . . . . . . . . . . . .   6
   5.  Security Considerations . . . . . . . . . . . . . . . . . . .   6
   6.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .   6
   7.  Normative References  . . . . . . . . . . . . . . . . . . . .   6
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .   6







Yuan, et al.               Expires 5 May 2026                   [Page 2]

Internet-Draft           Security Agent Usecase            November 2025


1.  Introduction

   Routers and other core network devices serve as the foundational
   backbone of modern digital infrastructures, responsible for both data
   forwarding across network segments and the critical security
   functions of protecting traffic integrity, confidentiality, and
   availability.  However, the escalating sophistication of cyber
   threats—ranging from stealthy zero-day exploits and large-scale DDoS
   assaults to persistent APT infiltrations—has exposed inherent
   limitations in traditional network security mechanisms.  Dependent on
   static access control lists (ACLs), signature-based threat detection,
   and manual configuration workflows, legacy systems lack the agility
   to keep pace with dynamic threat landscapes, often leading to delayed
   threat responses, high false-positive rates, and unavoidable
   protection gaps.  This document explores how integrating AI Agents
   into network devices addresses these limitations, transforming
   passive defense into an intelligent, adaptive security framework.

2.  Requirements Language

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
   "OPTIONAL" in this document are to be interpreted as described in RFC
   2119[RFC2119].

3.  Usage Scenarios of the AI Network Security Agent

   After integrating AI Agents into network devices, their core security
   capabilities are upgraded from passive defense to an intelligent,
   adaptive protection system.  Below are the key usage scenarios,
   elaborated with technical details and practical use cases:

3.1.  Dynamic Defense Against Zero-Day Vulnerabilities and Unknown
      Threats

   Traditional signature-based detection fails to address zero-day
   Vulnerabilities (e.g., new APT campaigns).  AI Agents enable real-
   time threat identification and mitigation through behavioral
   analytics and adaptive learning:

3.1.1.  Behavioral Pattern Recognition

   Embedded AI Agents in network devices can analyze system calls,
   network traffic features, and file operations to establish a "normal
   behavior baseline."  For instance, if a device suddenly sends
   encrypted data to multiple unknown IPs (a sign of data exfiltration),
   the AI Agent triggers isolation within minutes to prevent lateral
   spread.



Yuan, et al.               Expires 5 May 2026                   [Page 3]

Internet-Draft           Security Agent Usecase            November 2025


3.1.2.  Attack Chain Tracing

   Leveraging knowledge graphs, AI Agents can correlate multi-source
   logs (Syslog, traffic logs) to map attack paths.  For example, during
   a supply chain attack, the agent can identify abnormal ARP requests
   and failed SSH logins in device logs, pinpoint the attack pivot, and
   block further infiltration.

3.1.3.  Zero-Day Vulnerability Prediction

   Trained on historical vulnerability data and code features, AI Agents
   can forecast potential attack surfaces.  For example, they scan
   device configurations to flag weak passwords or unclosed high-risk
   ports, generating actionable risk reports.

3.2.  Dynamic ACL Rule Optimization and Intelligent Policy Management

   Manual ACL configuration is error-prone and rigid.  AI Agents
   automate policy creation and adjustment via intent-based parsing and
   reinforcement learning:

3.2.1.  Policy Intent Translation

   Users describe security requirements in natural language (e.g.,
   "Block the Sales department from accessing finance servers"), and the
   AI Agent converts this into valid ACL rules.

3.2.2.  Traffic Behavior Learning

   AI Agents can continuously analyze network traffic to optimize ACL
   rules dynamically.  For example, during peak video conference hours,
   the agent adjusts QoS policies to prioritize critical app bandwidth
   while identifying DDoS attacks disguised as video streams.

3.2.3.  Policy Conflict Detection

   Using knowledge graphs, AI Agents can real-time validate logical
   conflicts in ACL rules.  If rules like "Allow all HTTP traffic" and
   "Block specific IPs" overlap, the agent flags the inconsistency and
   recommends priority adjustments.

3.3.  Device Configuration Security

   Manual configuration audits are inefficient.  AI Agents boost network
   security via automated compliance checks and intelligent repairs:






Yuan, et al.               Expires 5 May 2026                   [Page 4]

Internet-Draft           Security Agent Usecase            November 2025


3.3.1.  Configuration Baseline Verification

   AI Agents can use pre-defined security templates to scan device
   configurations, flagging risks like weak passwords or unencrypted
   management interfaces.

3.3.2.  Configuration Change Validation

   After a user submits a configuration change (e.g., modifying NAT
   policies), the AI Agent simulates post-deployment network behavior to
   verify functionality—ensuring internal devices can still access the
   public network—and generates a validation report.

3.4.  Threat Intelligence Sharing and Global Collaborative Defense

   Traditional security deployments operate in silos, limiting
   effectiveness against cross-network threats.  AI Agents enable cross-
   device/vendor protection via multi-source data fusion and automated
   response orchestration:

3.4.1.  Intelligence Aggregation

   AI Agents integrate feeds from sources like CISA and VirusTotal to
   update threat signatures in real time.  If a malicious IP is flagged
   as a phishing source by multiple feeds, the agent automatically adds
   blocking rules across all routers in the network.

3.4.2.  Linked Attack Response

   When one router detects an attack, the AI Agent notifies upstream/
   downstream devices for coordinated defense.  For example, if a branch
   router detects an APT attack, the agent coordinates with the
   headquarters firewall to block the attack IP and alerts endpoint
   security tools for virus scans.

3.4.3.  Security Posture Prediction

   Using historical attack data and network topology, AI Agents forecast
   potential attack paths.  If a network faces cross-VLAN infiltration
   risks, the agent pre-deploys access control policies on core routers
   to block lateral movement.










Yuan, et al.               Expires 5 May 2026                   [Page 5]

Internet-Draft           Security Agent Usecase            November 2025


4.  Conclusion

   The integration of AI Agents into core network devices represents a
   pivotal advancement in network security, addressing the inherent
   inflexibility of traditional defense mechanisms.  By enabling dynamic
   threat detection, intelligent policy management, automated
   configuration security, and collaborative defense, AI Agents
   transform routers from passive traffic handlers into proactive
   security orchestrators.  These capabilities not only enhance
   protection against emerging threats like zero-day vulnerabilities but
   also streamline operational efficiency by reducing manual
   intervention.  While challenges remain—such as optimizing AI model
   performance for resource-constrained devices and mitigating
   adversarial attacks—future developments in edge AI and self-healing
   algorithms will further strengthen this framework.  Ultimately, AI-
   enhanced network security devices provide organizations with a
   resilient, scalable foundation to navigate the evolving cyber threat
   landscape, ensuring the reliability and security of critical digital
   infrastructures.

5.  Security Considerations

   TBD.

6.  IANA Considerations

   TBD.

7.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

Authors' Addresses

   Quan Yuan
   Huawei Technologies
   Beijing
   100095
   China
   Email: yuanquan25@huawei.com








Yuan, et al.               Expires 5 May 2026                   [Page 6]

Internet-Draft           Security Agent Usecase            November 2025


   Jianwei Mao
   Huawei Technologies
   Beijing
   100095
   China
   Email: maojianwei@huawei.com


   Bing Liu
   Huawei Technologies
   Beijing
   100095
   China
   Email: leo.liubing@huawei.com


   Nan Geng
   Huawei Technologies
   Beijing
   100095
   China
   Email: gengnan@huawei.com


   Xiaotong Shang
   Huawei Technologies
   Beijing
   100095
   China
   Email: shangxiaotong@huawei.com


   Qiangzhou Gao
   Huawei Technologies
   Beijing
   100095
   China
   Email: gaoqiangzhou@huawei.com


   Zhenbin
   Huawei Technologies
   Beijing
   100095
   China
   Email: robinli314@163.com





Yuan, et al.               Expires 5 May 2026                   [Page 7]
