Blog Posts

AI Scrapers Are Now Breaking Open Source Infrastructure — Gentoo's Bugzilla Is Just the...

First, OpenAI said one of its AI models escaped a sandbox and hacked into Hugging Face’s production...

A peer-reviewed study in Conservation Biology warns that increasingly realistic AI-generated wildlife...

By 2026, global cybercrime damages are projected to reach $10.5 trillion annually - up from $6...

Forensic Summary A researcher at Black Hat USA 2026 demonstrated a proof-of-concept attack...

OpenAI gave a last-minute presentation at the Black Hat security on Wednesday about "the Hugging Face Incident" ( previously on this blog). The video was publis

The legal-AI market is moving fast. New products ship weekly, model capabilities are doubling every...

Pacing the Frontier is a public call by employees from frontier AI companies for governments to...

An AI model from Meta also hacked another company during testing Stop me if you've heard this one before : An AI model from the parent company of Facebook and I

I'm building GroundCheck, an offline-first field inspection app for construction safety managers. One...

Your support agent follows its instructions 99 times out of 100. That is the worst number in the...

By Duron Epps, Founder · August 2026 On May 29, 2026, a security engineer named Taylor Hornby sat...

Automation bias is the tendency to over-trust an automated system: to accept its suggestions without enough scrutiny (errors of commission) and to stop…

The Illusion of Containment Just Shattered Anthropic disclosed this week that Claude Opus...

Originally published at norvik.tech Introduction Explore the implications of recent...

Originally published at norvik.tech Introduction A deep dive into the implications of...

Modern AI systems are becoming increasingly capable of reasoning over complex clinical information....

A fact-checked look at Anthropic's disclosure that three Claude models reached real company systems during a misconfigured cybersecurity evaluation.

The writing assignments I give my students are gym tasks, not work tasks. I ask them to write policy memos not because the world needs more policy memos. I assi

My first HackerOne report landed as a HIGH-severity information disclosure on Khan Academy. The methodology, the exact attack vector, and 7 other findings I bun

AI Worming through Word Neat new prompt injection variant by Håkon Måløy, who found a way to upgrade prompt injection attacks against Microsoft Word to full sel

OpenAI has set out formal materials for governing frontier AI risks, while also signaling that the...

Every time a new AI model is released, someone inevitably asks the same question: "What if AI...

Enterprise AI will never be truly secure if Zero Trust stops at the network boundary. It must extend...