AI security

AI data leaks, breaches and research on how AI can be attacked, and what to do about it.

An AI app left 144,000 chat logs open online. What you type into AI isn't always private

Researchers found millions of files from the AI service 302.ai, including user chats with personal details, in an unprotected online storage bucket. Here's what to keep out of AI chats.

28 Oct 2026

Saved your AI API key in a coding tool? A breach is a reminder to protect it like a password

AI coding tool OpenCode says an intruder may have exposed OpenAI, Anthropic and Google API keys saved in its online console. Some users report surprise bills. How students and developers should handle AI keys.

28 Oct 2026

AI browsers can be tricked by hidden instructions on web pages

Researchers keep finding ways to hijack AI browsers and assistants that act for you, and attackers have started hiding instructions on real websites. Here's how to use AI agents more safely.

28 Oct 2026