OpenAI builds a kill switch for its research models

A security news feed covering active exploitation, major vulnerabilities, ransomware and data breaches, cloud and supply-chain attacks, AI-agent security risks, critical infrastructure incidents, and cybercrime campaigns. Notable items include Chrome and OpenSSH zero-days or root vulnerabilities, Barracuda gateway flaws, AWS integration backdoors, ransomware attacks against universities and libraries, large-scale customer data exposures, attacks affecting industrial and healthcare environments, and emerging risks from autonomous AI agents.