Security vulnerabilities spread across popular AI tools, with two also found in Microsoft Copilot

Security firms found two vulnerabilities in Microsoft 365 Copilot that could let hackers trick the AI into exposing customer files and emails.

The first flaw was reported by Rubrik in April and has since been patched, while patch status for the second issue remains unclear.

Because Copilot can access enterprise work data, exploitation could lead to large-scale leaks of personal and business documents.

The case highlights how security in AI productivity tools directly affects product trust and enterprise adoption.