https://trydoberman.dev/blog/ Field notes from building Doberman: AI agent security, runtime authorization, and lessons from real incidents. en https://trydoberman.dev/blog/pacing-the-frontier-is-not-enough/ https://trydoberman.dev/blog/pacing-the-frontier-is-not-enough/ Tue, 15 Sep 2026 23:46:48 +0000 Dario Amodei’s We Must Pace the Frontier makes a strong case that AI capability is beginning to move faster than our ability to understand and control it. https://trydoberman.dev/blog/when-my-guardrail-crashes-the-agent/ https://trydoberman.dev/blog/when-my-guardrail-crashes-the-agent/ Thu, 10 Sep 2026 19:29:14 +0000 The easiest way past most security checks is to make them fall over. I built Doberman so that its own failure is a denial, and that choice has a cost worth being honest about. https://trydoberman.dev/blog/what-it-cost-me-to-cut-the-third/ https://trydoberman.dev/blog/what-it-cost-me-to-cut-the-third/ Wed, 09 Sep 2026 19:27:22 +0000 Simon Willison’s rule says an agent can’t safely hold private data, read untrusted content, and have a way out. A coding agent needs all three, so I put a check on the third one. This is the bill. https://trydoberman.dev/blog/approval-fatigue-is-a-security-issue/ https://trydoberman.dev/blog/approval-fatigue-is-a-security-issue/ Fri, 28 Aug 2026 17:02:19 +0000 I’ve spent the last two weeks making Doberman ask for approval less often. This increased safety and here’s why. https://trydoberman.dev/blog/cisco-defenseclaw-vs-doberman/ https://trydoberman.dev/blog/cisco-defenseclaw-vs-doberman/ Sat, 08 Aug 2026 23:34:11 +0000 Cisco's answer to runtime AI defense — what DefenseClaw gets right, and where a dedicated authorization layer makes a different bet. https://trydoberman.dev/blog/ai-security-playbook/ https://trydoberman.dev/blog/ai-security-playbook/ Thu, 06 Aug 2026 15:11:32 +0000 Takeaways from RAND's Practical Guide for Securing AI Models — and why AI security is shifting from model safety to execution security. https://trydoberman.dev/blog/when-ai-output-stops-being-text/ https://trydoberman.dev/blog/when-ai-output-stops-being-text/ Mon, 03 Aug 2026 01:26:29 +0000 The Hugging Face incident shows why agent security cannot end with better prompts, safer models or stronger sandboxes.