Spammers exploit ASCII smuggling to bypass AI email filters
Spammers are reportedly using invisible Unicode characters—a technique called ASCII smuggling—to bypass AI-powered email filters and conduct phishing campaigns. Microsoft detected a spike in such attacks in early 2026, with detection signatures rising to 2.5 million over four days, suggesting the technique is being widely adopted to evade security systems.
What happened
Spammers are using invisible Unicode tag characters—a technique called ASCII smuggling—to evade email filters and conduct phishing campaigns. Microsoft researchers observed a high-volume phishing campaign in early 2026 where ASCII smuggling signatures detected by Microsoft Defender for Office spiked from roughly 21,000 per day in February to more than 1.3 million, rising to 2.5 million over four days before falling sharply in mid-May.
The technique exploits the property that tag characters are invisible to humans but exist at the text-processing level, allowing spammers to obfuscate keywords before detection systems evaluate them. The same mechanism has previously been used for prompt injection attacks on AI agents.
Reported impact
Email spam delivered via ASCII smuggling surged significantly in 2026, with 2.37 million phishing emails at peak volume. Police officials reported similar threat emails from different parts of the country, suggesting a broader pattern of attack.
Index analysis
As AI-powered email filters have become more sophisticated, spammers have adapted by repurposing AI-era evasion techniques to improve traditional phishing tactics. The spike in ASCII smuggling detections demonstrates that spam detection remains an ongoing cybersecurity challenge even with modern AI defences in place.
Governance controls
Microsoft provided recommendations to developers to re-programme email filters in response to ASCII smuggling attacks.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.