AI "Mind Viruses" Can Spread Between Agents Through Persistent Prompt Files

The feed highlights active exploitation of critical vulnerabilities, credential theft, malware and botnet campaigns, supply-chain and typosquatting attacks, cloud and SaaS abuse, AI-agent prompt persistence risks, espionage, ransomware, and phishing. Multiple incidents involve unauthenticated remote code execution or high-impact data compromise, including GitLab, Forminator, VMware vCenter, SAP Commerce Cloud, Apple macOS Screen Sharing, and Ray. Threat activity spans nation-state actors, criminal groups, and opportunistic campaigns targeting enterprise infrastructure, developers, browsers, Io