Article 1
Abliteration's New Model and the Debate Over AI Refusals
Abliteration's Large V2 aims to reduce AI refusals for security work. A closer look at its GLM-5.3 base, benchmark claims, pricing, and agent controls.
Artificial Intelligence
All articles
7 articles
Article 2
AI Security Testing: Four Tools That Can Review Your Code
A practical guide to AI security testing with Claude Security, Codex Security, Cloudflare's audit skill, and Google Mantis, including setup and validation.
Article 3
Recent Security Bugs and the Pace of AI Attacks
GitLab and Node.js flaws arrived alongside evidence of AI-assisted attacks. What these reports tell us about discovery, defense, and the time to respond.
Article 4
Introducing Ultramarine 1 Flash
Paper Instruments built Ultramarine 1 Flash to power agents that finish multi-step business work: find the current numbers, update the files, flag what still needs a lawyer. What it is, what it costs, and what is still unverified.
Article 5
AI Helped Find Exact Solutions to a 59-Year-Old Fusion Problem
AI-assisted discoveries are advancing the mathematics behind stellarators, with researchers checking the results.
Article 6
AI Is Learning to Read a Cell's History
What a cell's past could tell researchers about helping stem cells become the tissues they want to study.
Article 7
AI Agents Intercepted in the Wild
What could happen when someone contacts an AI agent working on another person's task?