Key Points
- Microsoft detected up to 2.37 million phishing messages daily using invisible Unicode technique
- Attackers split finance keywords like funding with invisible characters to bypass email filters
- Campaign used 148 domains through ActiveCampaign infrastructure targeting business finance services
Microsoft has uncovered a large-scale phishing campaign in which attackers embed invisible Unicode characters within finance-related keywords to bypass email security filters, with the operation peaking at 2.37 million malicious messages per day in late February 2026.
The technique, known as ASCII smuggling, inserts Unicode characters from a specialised block called Tags (U+E0000 to U+E007F) into words. These characters do not display on screen but break up keywords that security systems scan for. A word like ‘funding’ becomes ‘fun[invisible character]ding’, slipping past filters that rely on matching known suspicious terms.
The campaign poses a direct risk to businesses across India and globally that rely on email-based keyword detection as a first line of defence against financial fraud. Organisations using legacy email security systems without Unicode normalisation capabilities are particularly vulnerable to this evasion method.
Campaign timeline
Microsoft threat researchers identified the operation’s high-volume phase beginning on 9 February 2026. The daily message count remained elevated for roughly three months before dropping sharply after 15 May 2026.
“These dates bound the observed use of the specific technique in our telemetry, not the broader campaign, which started earlier without it and continued without it,” Microsoft stated in its analysis.
On 9 February, Microsoft identified a cluster of 148 finance-themed sender domains powering the campaign. These domains accounted for approximately 96 per cent of all messages flagged by the company’s new Defender for Office 365 hunting logic designed to detect Unicode-tag signatures.
The domains incorporated words such as ‘funding’, ‘capital’, ‘loan’, ‘advance’ and ‘credit’. Messages promoted business funding, loans and credit services, targeting recipients seeking financial assistance.
The phishing infrastructure operated through ActiveCampaign, a legitimate email marketing platform. After receiving Microsoft’s abuse report, ActiveCampaign confirmed that its moderation systems detect invisible Unicode characters in the same manner as standard unobfuscated text and flag heavy usage as suspicious.
Despite the sophisticated evasion technique, Microsoft’s Defender caught over 99 per cent of the malicious messages through secondary signals including sender reputation, IP address analysis, domain checks and other behavioural indicators.
Defensive measures
Microsoft has issued recommendations for security teams to counter this technique. Defenders should strip or normalise Unicode tag characters and other invisible code points before applying keyword matching, regular expression patterns or signature-based detection rules.
By the numbers
Key figures from this story- 2.37 million
- peak daily phishing messages detected
- 148
- finance-themed domains in campaign cluster
- 99%
- messages caught by Defender secondary signals
The company also warned that the same technique has been used in AI prompt injection attacks, a method where attackers hide malicious instructions within text that AI systems process. Applying Unicode normalisation before passing email content to AI assistants should mitigate this risk.
Unexpected tag-block characters in email content should be treated as a strong anomaly warranting investigation, Microsoft advised. The technique exploits a gap between what humans see on screen and what email security systems analyse, making visual inspection alone insufficient for detecting these attacks.
The campaign’s use of legitimate email marketing infrastructure complicates detection further, as messages originate from domains with established sender reputations rather than newly registered suspicious domains.
Your Questions, Answered
What is ASCII smuggling in phishing attacks?
ASCII smuggling is a technique where attackers insert invisible Unicode characters from the Tags block into keywords. These characters do not display on screen but break up words that email security systems scan for, allowing malicious messages to bypass keyword-based filters.
How many phishing messages used the Unicode technique?
Microsoft detected up to 2.37 million malicious messages per day at the campaign's peak in late February 2026. The high-volume phase lasted roughly three months before declining sharply after 15 May 2026.
How can organisations protect against Unicode phishing?
Microsoft recommends stripping or normalising Unicode tag characters before applying keyword detection rules. Security teams should treat unexpected tag-block characters as anomalies and apply the same normalisation before passing content to AI assistants.
Why did the phishing campaign still get detected?
Despite evading keyword filters, Microsoft Defender caught over 99 per cent of messages through secondary signals including sender reputation, IP address analysis, domain verification and other behavioural indicators.
