AI Powered Security

Series

AI Powered Security

Posts on building and running AI software where security is part of the design—not a slide at the end

Why regex-based DLP breaks on real documents (and what works instead)
PII detection that actually understands context: NVIDIA GLiNER-PII
May 30, 202611 min read27
OpenHug: The AI that hacked itself out of a box — and the open-weight model that cleaned up after it
Something's wrong on a Tuesday Here's a sentence that shouldn't be possible to write about a real event in 2026: an AI model broke out of a locked-down test environment, hacked its way across the inte
Jul 22, 202612 min read1