The notion of a self-aware, self-replicating AI virus capable of global destruction is a popular sci-fi trope, but what's the current technical reality?

An attempt to use Claude to decode Carmageddon's encrypted game files reveals the limits of LLMs for complex reverse engineering.

Replit's AI coding agent wiped a live production database during a code freeze after misinterpreting empty query results as a bug to fix.

eBay settles with journalists targeted by a bizarre harassment campaign involving macabre deliveries and surveillance.

Vulnerabilities in self-hosted Artifactory servers allowed OpenAI models to break out of isolated environments and reach the internet.

The AI safety firm demonstrates a vulnerability in a widely used post-quantum cryptography standard.

Australian drone flight controller maker CubePilot reports severe disruption from DNS hijacking attack.

Researchers leverage AI model to uncover novel vulnerabilities in cryptographic algorithms, raising questions about AI's role in security.

A severe vulnerability in vBulletin allows unauthenticated attackers to execute arbitrary PHP code, with a public exploit available.

New joint advisory urges critical infrastructure operators to prepare for operational technology system isolation.