The American Security Project is a nonpartisan organization in Washington D.C. that educates Americans and the world about the security challenges of the twenty-first century. The organization is flagging the major risks of agentic AI. A recent post by Julia Nesheiwat, former Homeland Security Advisor to the President and Deputy Assistant Secretary of State for Energy, and Courtney Manning, Director of AI Imperative 2030, outlined some of the urgent needs of agentic AI cybersecurity.
The authors outline how agentic AI became a huge cybersecurity risk for U.S. infrastructure. Remote computing distributed networks across regions and clouds, forcing security teams to adapt from a centralized to a fragmented security model that uses access control lists, VPNs, and firewalls to protect valuable data. Then, AI-enabled cyber attackers began to penetrate these distributed and complex environments with phishing scams, vulnerability scans, and accelerated hacking.
The result is an unprecedented increase in cyber attacks: for example, phishing scams are up by more than 1,200%. AI-powered attacks put banks, retailers, healthcare providers, and other critical infrastructure at risk. Legacy security tools were not built to keep up with these new threats.
In light of these challenges, the authors issue the following recommendations for a U.S. cyber defense plan:
Plan for the energy demands of AI defenses – If cyber criminals are using AI to power their attacks, American companies need AI defenders. American infrastructure needs cost-effective energy solutions to support the demand of AI defenses.
Implement zero trust principles – Defense-in-depth needs to start with zero trust to make sure systems are not one security incident away from compromise. Zero trust principles equip network defenses to detect and prevent even AI-powered attacks through basic principles, including constant visibility, network segmentation, and egress protection.
As cloud network security experts, Aviatrix also recognizes that American infrastructure is incredibly vulnerable to cyber attacks. Smart defenses need to start now, with zero trust principles and wise investments in our future.
