Writing
Essays & field notes.
Jul 24, 2026
Your AI Didn't Go Rogue. It Used the Access You Gave It.
Most enterprise AI incidents this year weren't the model going off-script. They were the model using access it should never have had. A field report on the over-privileged agent problem.
Jul 2, 2026
What Enterprises Actually Get Wrong When They Deploy AI
A field report on the recurring ways enterprises get AI security wrong, and the mostly-familiar discipline that closes the gap.
May 31, 2026
Building an AI Security Lab You Can Actually Run
Most AI security content stops at the definition. This is a hands-on, fully local lab for the attack surface that actually matters — where untrusted input meets real capability.