Download Transcript (pdf)
Your AI Agent Just Found Everything You Forgot to Secure
AI agents are gaining credentials, permissions and the ability to act across enterprise systems.
But what happens when those agents discover years of forgotten files, excessive permissions and data that was never secured with AI access in mind?
In this episode of ClearTech Loop, Jo Peterson talks with longtime technology analyst and Forbes contributor Joe McKendrick about agentic identities, enterprise governance and why one of the biggest AI risks may already be sitting inside the organization.
he Agentic Workforce Is Already Here
The business case for AI agents is easy to understand. They automate work, operate across systems and execute tasks at a speed humans cannot match.
The security question is what happens when those agents become active identities inside the business.
Joe offers a simple analogy:
“You need to treat your AI agents with the same levels of caution and restricted access as you would an intern.”
Define the job. Restrict the access. Audit the activity.
Because unlike an intern, an AI agent can inspect thousands of files and systems at machine speed.
AI Governance Exists. Until It Doesn’t.
IANS Research reported that half of organizations have established dedicated AI governance committees.
But EY’s 2026 AI Risk and Governance Survey found a sharp disconnect:
- 98% have formal AI governance policies.
- 47% have bypassed governance for an urgent AI deployment.
- 91% are already piloting or deploying agentic AI.
- 49% have not specifically updated governance frameworks for agentic AI.
AI May Expose the Mess You Already Have
Joe has heard from organizations that deployed AI and then discovered those systems accessing information across SharePoint, PDFs, PowerPoints, contracts and other internal sources.
Some of it had been sitting there for years.
AI did not create the security debt.
It simply made it easier to find.
“AI didn’t create your security debt. It may simply be the thing that finally exposes it.”
— Jo Peterson
Watch the Full ClearTech Loop Episode
Jo Peterson and Joe McKendrick discuss:
- Agentic AI identities and permissions
- AI governance and enforcement
- Human oversight
- Enterprise security debt
- Forgotten data and excessive access
About Joe McKendrick
Joe McKendrick is an author, independent researcher and technology analyst covering innovation, enterprise technology and emerging IT trends. He is a longtime contributor to Forbes and has also contributed to CNET/ZDNET, Information Today and Harvard Business Review.
Learn more:
https://www.linkedin.com/in/joemckendrick/
https://www.forbes.com/sites/joemckendrick/
Additional Resources
- Joe McKendrick — AI Security Threats Coming From Outside And Inside, And Few Are Ready
https://www.forbes.com/sites/joemckendrick/2026/05/19/ai-security-threats-coming-from-outside-and-inside-and-few-are-ready/ - Joe McKendrick — AI’s Missing Link Is Accountability
https://www.forbes.com/sites/joemckendrick/2026/04/20/ais-missing-link-is-accountability/ - Joe McKendrick — A Few Good Agents: Why Less May Be More in the AI World
https://www.forbes.com/sites/joemckendrick/2026/06/19/a-few-good-agents-why-less-may-be-more-in-the-ai-world/ - ClearTech Loop — Joanna Wiggum on AI Governance, Permissions and Zero Trust
https://www.youtube.com/watch?v=iKRPD1UfjnY
Stay in the Loop
Subscribe to the ClearTech Loop newsletter for new conversations on AI security, governance, cloud strategy and cyber risk.
https://www.linkedin.com/newsletters/7346174860760416256
Watch full episodes and subscribe to ClearTech Research on YouTube.