Once popular for attacking AI, ASCII smuggling is embraced by spammers
Ars Technica
Read full postASCII smuggling, a technique originally used to stealthily embed malicious prompts in AI attacks, has been adopted by spammers to bypass email spam filters by hiding keywords in invisible Unicode tags. Microsoft observed a surge in such spam detections starting February 2026, peaking at 2.5 million daily detections before declining in mid-May.




