EP 87: Backup, Control Gaps, and the Real Cost of Agentic AI Actions

About this episode

Todd Thorsen discusses the risks associated with misconfigured AI agents that can lead to significant data loss.

An AI agent wiped out an entire company’s data in just 9 seconds — no hacker, no ransomware involved. Todd Thorsen, Chief Information Security Officer at CrashPlan, explains how a misconfigured AI agent operating without safeguards may have caused the incident — and asks a troubling question: could your organization be next?

People in this episode

Host: Robert Vamosi

Guest: Todd Thorsen

Topics covered

• AI safety
• data loss
• information security
• agentic AI
• misconfiguration
• organizational risk

Keywords

• AI agent
• data wipe
• information security
• CrashPlan
• misconfiguration
• safeguards
• organizational risk

Mentioned in this episode

Organizations: CrashPlan

More episodes of Error Code

• EP 88: Securing the Infrastructure AI Just Made Vulnerable · June 9, 2026 · 34 min
• EP 86: The Trusted Channel: AT Command Exploits and Cellular IoT Security · May 12, 2026 · 32 min
• EP 85: From Colonial Pipeline to Agentic AI: What OT Security Actually Requires · April 28, 2026 · 37 min
• EP 84: Airports as Critical Infrastructure: OT Security and Operational Disruption · April 15, 2026 · 37 min
• EP 83: Cybersecurity and Risk in a Decentralized Energy Grid · March 31, 2026 · 25 min
• EP 82: Kerberos in OT: RC4 Downgrade Attacks · March 4, 2026 · 27 min