Microsoft boosts AI‑driven Windows security as cyber attacks target its services

Microsoft has introduced an AI system called MDASH (Multi‑model Agentic Scanning Harness) to examine Windows source code. The tool uses more than one hundred specialized AI agents to locate vulnerabilities, reducing the time between discovery and patching. In May 2026 it identified 16 previously unknown flaws, and the June 2026 Patch Tuesday recorded over 200 fixes, meaning users can expect larger, more frequent security updates.

Security researchers have warned of a surge in vishing campaigns aimed at Microsoft 365 users. Attackers place phone calls, impersonate support staff and direct victims to counterfeit Microsoft Entra ID login pages, seeking to steal credentials and gain access to corporate accounts.

A separate threat group, identified as O‑UNC‑066 (also called Pink), is hijacking Passkey registration processes. Through voice‑phishing and a manually operated PHP phishing kit, the group registers attacker‑controlled FIDO2 keys in victim accounts, affecting six industry sectors and using the stolen credentials for data extortion.

Doctor Web uncovered a multi‑stage malware that injects malicious code into Visual Studio C++/C# projects. The tampered projects propagate the payload into every binary built with the compromised environment, creating a supply‑chain infection vector.

Microsoft also disclosed a new destructive Windows malware named GigaWiper. One variant wipes hard drives completely, while a second adds a persistent backdoor and ransomware capabilities, encrypting files with unrecoverable keys and erasing system logs.