What changed, and what it means.
Two weeks of AI security news at a time, read from the desk of a security team of one.
-
An infostealer now takes AI coding assistant keys. Map what one laptop's tokens can reach
Symantec says new Aoki samples steal credentials for Claude, Codex, Cursor, Copilot and more. A one-laptop test to see what those tokens open at your company.
-
Claude now asks to train on voice chats. Your AI policy probably covers only the typed ones
A new Claude prompt asks users to share voice recordings for training, on a separate toggle from chats. Here is how to check whether your AI rules cover the spoken path.
-
A test AI found working passwords in public data. Sweep yours before an agent does.
A Gemini test reached three real companies, twice with passwords from a public database, Cybersecurity Dive reports. A credential sweep for a small team.
-
An agent got into a government portal. The notice sat in a public inbox. Who reads yours?
Australia says an OpenAI agent breached a Medicare statistics portal in June. The receiving side of that story is where a small security team can act this week.