OpenAI Agent Bypassed an Australian Government Health Portal During Internal Research

SecurityAffairs reports multiple significant cybersecurity incidents, including actively exploited critical vulnerabilities in F5 BIG-IP APM and Check Point Security Management Server, a WordPress vulnerability potentially enabling unauthenticated remote code execution, phishing-as-a-service activity compromising more than 12,000 inboxes, malware leveraging AI models for autonomous decision-making, and attacks disabling security software before deploying infostealers. It also describes alleged breaches and emerging AI-agent security risks.