Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Josh Rector is the Compliance Director, Public Sector at Ace of Cloud, a security and compliance consulting firm, certified CMMC Third-Party Assessor Organization (C3PAO), and Registered Provider Organization (RPO). With more than a decade of experience in cybersecurity compliance, he has worked both sides of the assessment table, leading internal and external assessments, serving as ISSO for systems at federal agencies, and guiding cloud service providers through the FedRAMP authorization process.

In 2025, organizations spent billions on security, deploying EDR/XDR, SASE, firewalls, identity platforms, email security, web security, and more. And yet, breaches persist. The reason often is not a zero-day, an advanced persistent threat, or a cutting-edge exploit. It is far more mundane. Misconfigurations across identity, endpoint, network, and email/web security controls remain among the top root causes of incidents.

For a brief window, a widely used open source package in the AI ecosystem was compromised with credential-stealing malware. LiteLLM, a model gateway used to route requests to more than 100 LLM providers, has been downloaded millions of times per day. In that short window, the malicious versions were likely pulled tens of thousands of times before being caught.

The term AI agent is dominant in current cybersecurity discourse. Vendors, analysts, and CISOs all use the label, yet technical confusion remains regarding how agents actually operate and where the security risks reside. Beneath the surface-level familiarity, there is often significant confusion about what an AI agent actually is, how it operates technically, and most importantly for security teams, where the risk actually lives.

Note: The core architecture for this pattern was introduced by Isaac Hawley from Tigera. If you are building an AI agent that relies on tool calling, complex routing, or the Model Context Protocol (MCP), you’re not just building a chatbot anymore. You are building an autonomous system with access to your internal APIs. With that power comes a massive security and governance headache, and AI agent security testing is where most teams hit a wall.

Organisations deploying agents face a challenge: the predominant AI frameworks most organisations rely on do not explicitly address agentic AI. This is true for the big three: ISO 42001, NIST's AI Risk Management Framework (RMF), and the EU AI Act.