◢ DEV LOG

Articles

Notes on AI, development and visibility

prompt injectioncybersecurityartificial intelligenceai agentsai security

Prompt injection: what it is and how to protect your company

Prompt injection is an attack that hides instructions inside text an AI is going to read, like an email, a web page or a PDF, so it does something you never asked for. OWASP ranks it as the number one risk for AI applications, and it affects any company using AI assistants or agents.

October 12, 2026 · 6 min readRead article →
claudeanthropicclaude downclaude codeclaude apistatus

Claude is down: what's failing and what you can do

Claude, its API, Claude Code and Cowork went down on September 29, 2026 between 14:00 and 14:59 UTC. Service is back to normal. Here's what happened, how to check whether Claude is down and what to do next time.

September 29, 2026 · 3 min readRead article →
CybersecurityPhishingBusiness

Attacks don't come in through the server: they come in through an email

I've spent years stopping real attacks at an industrial company and almost none of them started on a machine: they started with a person. This is what every team should know so nobody becomes the way in.

September 17, 2026 · 4 min readRead article →
GEOSEOAILLM

What is GEO and how to get your website into ChatGPT, Perplexity and Gemini answers

GEO (Generative Engine Optimization) is structuring your website so AI engines cite it when they write answers. You're not chasing a spot in a list of links: you're chasing being the source the model picks. Here are the techniques that work in 2026.

July 8, 2026 · 3 min readRead article →