ASCII smuggling, a technique originally developed to hide prompt injections from AI models, is now being used by spammers to bypass email filters. The method relies on Unicode tags that are invisible to human readers but readable by machines. By inserting these tags into words, attackers can obfuscate keywords that security systems typically flag.
Microsoft reported a significant surge in the use of this technique earlier this year. In early February, detections by Microsoft Defender for Office rose from approximately 21,000 per day to over 1.3 million. Within four days, this figure reached 2.5 million. The high volume of spam persisted until mid-May, when it experienced a sharp decline.
Spammers utilize these hidden characters to disrupt the tokenization process in machine learning and natural language processing models. By breaking words into unexpected sub-tokens, the technique prevents filters from identifying common phishing terms like 'credit' or 'funding'. Microsoft has since issued guidance to developers on how to improve filter detection for these hidden tags.