Copy-paste is an untrusted input channel in agent systems. If your app can act through tools, paste becomes a security boundary.
AI Articles
Essays tagged AI by Jonathan R. Reed.
Notes on ai work
This archive groups writing by theme. Posts tagged AI include build notes, security lessons, design tradeoffs, and experiments that connect software decisions to real deployment constraints.
Use this page as a focused reading path through the site. The articles below are ordered newest first and link back into adjacent topics so you can move from a narrow subject into related AI, security, design, tooling, ethics, and development notes without relying on search alone.
Smaller tag archives are kept because they capture useful side threads across the work. Even when a tag only has a few posts, it gives readers a stable entry point into the ideas, references, and projects that shaped that part of the site.
The goal is not to turn every tag into a broad category page. It is to make the archive useful for people who arrive from search, a shared link, or an old project note and need enough context to know why the topic mattered here clearly before they keep reading.
The strongest posts in each archive are practical rather than broad: what broke, what changed, what I learned, and which constraint made the lesson worth saving.
Prompt injection is phishing for LLMs, and secure agent design depends on treating text as hostile input and tools as hard boundaries.
A practical method for evaluating stealth AI models, separating confirmed identity from inference, and recording evidence without trusting style or self-report.
The onslaught of AI news
2 minAI news is a firehose of fluff, broken publishing patterns, and missing feeds. I built an aggregator because checking each vendor site was worse.