Send us Fan Mail
AI is moving faster than most organizations can govern it.
Employees are already using multiple models. AI agents are gaining access to systems and data. Products are being built with AI embedded into core workflows. Yet many organizations still treat an AI policy as if it were an AI security program.
It isn’t.
In this episode of Musings from the Cyber Trench, I sit down with cybersecurity executive Jason Loomis to discuss what AI security looks like from the CISO seat—and why the hardest problems may have less to do with the technology than with leadership, accountability, access, and execution.
We get into the practical realities of securing AI and agentic systems, including:
- Why executives often have a very different picture of AI usage than what is actually happening inside the organization
- Why publishing an AI policy is only the first step toward meaningful governance
- How familiar security principles like MFA, least privilege, data inventory, and classification still apply to AI
- Why agentic AI creates new challenges around permissions, changing intent, task-scoped credentials, and time-bound access
- Which guardrails should be non-negotiable before AI can influence sensitive data, decisions, or customer outcomes
- Why startups and fast-moving companies often defer security—and the downstream cost of trying to bolt it on later
- Why executives remain accountable when AI makes or influences a bad decision
- How GRC teams can keep pace as AI frameworks, controls, products, and risks evolve faster than traditional governance cycles
- Whether AI can eventually monitor and govern other AI systems—and who watches the system doing the monitoring
- Why AI-generated software still requires human judgment, product thinking, security review, and an understanding of how people actually work
One theme keeps surfacing throughout the conversation: AI does not eliminate accountability.
If an AI agent has excessive permissions, leadership owns that risk. If an AI-enabled product creates an unacceptable outcome, leadership cannot simply say, “AI did it.” And if an organization waits until AI is already deeply embedded in production before bringing security into the conversation, remediation becomes significantly harder.
Jason’s advice to leaders is straightforward: start now.
You do not need to solve every AI security problem on day one. But you should know where AI is being used, understand what it can access, identify your highest risks, involve the people closest to the technology, and begin putting guardrails in place before an incident forces the conversation.
ABOUT JASON LOOMIS
Jason Loomis is an experienced cybersecurity executive and CISO whose background spans SaaS, healthcare, banking, finance, e-commerce, mergers and acquisitions, IPO and pre-IPO environments.
His work sits at the intersection of cybersecurity, organizational decision-making, risk, and leadership. In this conversation, Jason also discusses how his experiences outside cybersecurity—including disaster relief work and adventure sports—have influenced the way he thinks about leadership, resilience, and risk.
ZERO TRUST READINESS ASSESSMENT
Responsible for ICAM, Zero Trust, or identity security within a federal agency, prime contractor, or large regulated enterprise?
If you are trying to move from strategy to execution, start with Zephon’s Zero Trust Readiness Assessment:
https://zephon.tech/zt
Questions or guest suggestions?
defend@zephon.tech
Responsible for ICAM, Zero Trust, or identity security in a federal agency, prime, or large regulated enterprise?
If you’re trying to move from strategy to execution, start with Zephon’s Zero Trust Readiness Assessment: zephon.tech/zt
Questions or guest ideas? Email defend@zephon.tech