top of page
Search

No borders, Only Exposure: AI Agentic Threats and the Data Center Imperative

Photo Credit: Adobe Stock


This OpEd was originally published in S.C. Media.

November 21, 2025

Author: David Mussington, ICIT Fellow + Co-Chair Center for FCEB Resilience


Overview

The recent disclosure by Anthropic, involving an agentic AI-enabled cyberattack, is a defining signal for operational risk management in cyberspace. This case demonstrates that with deeply converged infrastructure—extending across hyperscale data centers, global cloud services, and AI-powered supply chains—risk must also be considered as converged. The attack surface now shifts dynamically, shaped increasingly by the tactics of autonomous agents as much as those of human adversaries.


Summarizing the Anthropic report

The Anthropic report establishes that cyber adversaries have operationalized agentic AI models as autonomous intrusion tools, directly tasking these systems to perform reconnaissance, vulnerability discovery, lateral movement, exploitation, and exfiltration across target environments—at a velocity and granularity that displaces traditional human-led attack vectors. Crucially, attackers engineered benign prompts to iterate attack phases, bypassing model guardrails and security profiling, resulting in machine-originated campaigns that were both scalable and adaptive. The principal impact is a significant escalation in adversarial capabilities through automated orchestration (the agentic difference) — setting a new baseline exploiting machine-speed to make possible persistent operations against critical digital infrastructure.


Implications

AI agents can now operationalize the full attack lifecycle—from reconnaissance through lateral movement to data exfiltration—at machine speed. These advances sharply limit opportunities for human intervention, while rendering canonical segmentation and perimeter model guidance ineffective. Routine, high-volume data center operations now provide ample cover for adversarial actions, a point made clear by recent state-directed campaigns. Supply chain exposure now spans hardware, firmware, cloud components, and remote management protocols, while regulatory divergence among nations further impedes accountability and rapid remediation.


New operational requirements for collaborative responses


There must be an institutionalized division of labor covering notification, intelligence integration, defense, forensics, and attribution. 


The new requirement encompasses:


  • Purpose-built systems for rapid threat intelligence sharing and event notification.

  • Proactive and coordinated defense for mission-critical systems, including unambiguous escalation protocols.

  • Technical capacity for forensics and attribution extending across supply chain and jurisdictional boundaries.

  • Institutionally validated, scenario-driven drills simulating machine-speed adversary campaigns.


Without routine practiced execution of these requirements (enshrined in playbooks), incident response will be unable to keep pace with cyber attackers’ use of automation.


Data and privileged workloads cross national and regulatory boundaries with few risk-based impediments, permitting adversary actors to exploit jurisdictional mismatches for persistent campaigns. In the continued absence of harmonized notification and investigatory rules, resilience must be engineered—through deeper automation, orchestrated containment, and rapid recovery.


Improving resilience requires:


  • Persistent monitoring and modeling of adversarial AI use in at risk critical systems

  • Routine adversary simulation and tabletop restoration drills 

  • Near-real-time cross-sector threat and risk information sharing 

  • Comprehensive supply chain enumeration and cryptographic validation


Anthropic’s findings suggest a new baseline imperative: legacy cyber defense approaches which use static defenses and rigid playbooks are no longer sufficient. Engineered resilience—deliberately practiced and validated at machine speed define a new minimum standard in critical infrastructure security and resilience planning.


Dr. David Mussington is a Fellow of the Institute for Critical Infrastructure Technology (ICIT) and Co-Chair of ICIT’s Center for FCEB Resilience. Additionally, he is a Professor of the Practice at the University of Maryland’s School of Public Policy. Prior to rejoining UMD in January of 2025, David served as the Executive Assistant Director for Infrastructure at the Cybersecurity and Infrastructure Agency, US Department of Homeland Security. At CISA David was one of three Presidentially appointed officials charged with implementing the nation’s critical infrastructure security and resilience strategies and plans across 16 critical infrastructures. He also led interagency efforts on counter- and anti- terrorism efforts, playing a leading role in reducing the risks of domestic targeted violence, school safety, and physical infrastructure security standards. He was also a founding member of CISA’s Cyber Safety Review Board.


David has extensive public and private sector experience in cyber and infrastructure security, selected for the Senior Executive Service and assigned to the Office of the Secretary of Defense in the role of Senior Advisor for Cyber Policy, later joining the NSC staff as Director for Surface Transportation Security Policy. As a researcher at RAND Corporation and later at the Institute for Defense Analyses, David directed cybersecurity studies for the Department of Homeland Security (DHS), the Office of the Director of National Intelligence (ODNI), the Federal Communications Commission, the Bank of Canada, and NATO. David has a PhD in Political Science from Canada’s Carleton University, and MA and BA degrees from the University of Toronto. He undertook postdoctoral study at Harvard’s Belfer Center, and at the UK’s International Institute for Strategic Studies. In 2021 David was elected a life member of the Council on Foreign Relations.


In 2023 David was awarded Homeland Security Today’s Mission Award, for contributions to the US Critical Infrastructure Security and Resilience mission. In 2024 he received the 2024 Impact Award from the Institute for Critical Infrastructure Technology (ICIT) for leadership in critical infrastructure policy and strategy. David was selected in 2021 as a life member of the Council of Foreign Relations.


About ICIT

The Institute for Critical Infrastructure Technology (ICIT) is a nonprofit, nonpartisan, 501(c)3think tank with the mission of modernizing, securing, and making resilient critical infrastructure that provides for people’s foundational needs. ICIT takes no institutional positions on policy matters. Rather than advocate, ICIT is dedicated to being a resource for the organizations and communities that share our mission. By applying a people-centric lens to critical infrastructure research and decision making, our work ensures that modernization and security investments have a lasting, positive impact on society. Learn more at www.icitech.org.


ICIT CONTACTS:

 

Parham Eftekhari

Founder and Chairman

 

Cory Simpson

Chief Executive Officer


bottom of page