Once popular for attacking AI, ASCII smuggling is embraced by spammers

A sophisticated technique known as ASCII smuggling, originally developed to bypass security filters in AI prompt injection attacks, is now being utilized by spammers to evade email security protocols. The method involves encoding malicious content using a specific range of Unicode tags that mimic standard ASCII characters. While these characters are easily processed and interpreted by computer systems and Large Language Models, they remain virtually invisible to human readers. By leveraging this discrepancy, attackers can embed hidden instructions or spam content that bypasses traditional text-based filtering systems. Originally gaining notoriety for making AI attacks more stealthy, the technique has evolved into a broader threat for email platforms. Security experts note that this shift highlights the ongoing cat-and-mouse game between attackers and automated defense systems, as malicious actors continue to repurpose advanced technical exploits to circumvent modern content moderation and anti-spam measures.