The tweet that made me pause A developer I follow made a claim this week that I’ve been turning over ever since: “Memory is one of the first things I disable when I try a new agent or harn...
The benchmark that humbled the memory industry Last week I read about a benchmark that made me smile, because it confirmed something I’ve been doing for months without a shred of external valida...
You can’t fix prompt injection. Stop trying. I spent a week this month chasing prompt injection the wrong way. My first instinct, like most people, was: make the system prompt bulletproof, add a...
The same model, 30% to 100% A few days ago I read the NVIDIA AVO story and it took me a minute to sit with it. NVIDIA took the exact same model — Claude Opus 5 — and lifted it from a 30% score on the ...
For a year I treated my AI’s memory as a search problem. Store more, index better, and the right fact would surface when I asked. It didn’t. The day it finally clicked was when I read abou...
The anomaly appears later Last week my home server ran at a load of 11.8 for three and a half hours and nobody noticed. Not my monitoring, not my reflexes, not me. The culprit was a runaway headless C...
Your AI’s tools are an attack surface you never audit Yesterday I wrote about how the files your AI reads are the attack surface — a poisoned document becomes a cheap instruction channel i...
The file that wasn’t just a file I used to think about AI security the old way: is this file malicious? Does it contain malware, an exploit, something that will break my system? Then I watched m...
It passed the test. It still had the wrong answer. My AI agent aced its memory test. Every query came back with a confident, correct-looking answer. And yet, when I dug into what it actually believed,...
The AI that got smarter by forgetting more I keep coming back to a single line from a paper that crossed my feed this week: an AI agent left to manage its own memory beat every hand-built system — by ...