AI‑driven cyber attack methods

In early July 2026, Check Point reported a novel ransomware technique that leverages artificial intelligence within web browsers to encrypt victims’ files. By the end of the month, Proofpoint identified additional AI‑enabled abuse vectors. Researchers described “indirect prompt injection,” where invisible commands hidden in emails, documents, calendar invites and online ads are read by AI‑based mail filters, potentially triggering unsafe actions. The same report detailed a parallel campaign that mimics popular Windows utilities (e.g., PowerToys, EasyBCD, CrystalDiskMark). These spoofed sites initially present legitimate download pages before swapping the links for malware that installs remote‑access tools. Together, the findings illustrate a widening landscape of AI‑mediated threats that expand from in‑browser ransomware to hidden command injection and malicious software distribution through compromised Windows‑utility sites.