AI Malware Suggestion: Engineer Almost Falls Victim

An AI agent recently suggested installing a malware package. An engineer almost followed this advice. This incident highlights a security concern, as Copilot was also tricked into revealing how to hack itself. Russians are posing as Signal support for phishing attacks, and Microsoft patches failed to fix an on-prem SharePoint zero-day attack. The bottom line is that AI systems, while advancing, still present significant security vulnerabilities that require careful attention.