Invisible Ink in the Inbox: How Phishers Borrowed an AI Trick to Flood Filters With Millions of Loan Lures

Phishers used invisible Unicode tag characters to split keywords like 'funding' in millions of loan-themed emails, bypassing filters that saw broken strings while recipients saw clean text. The campaign peaked at 2.37 million messages daily before declining, showing how AI-era tricks now fuel traditional spam at scale. Microsoft researchers uncovered the adaptation in early 2026.