Microsoft has issued an alert about a high-volume phishing campaign leveraging invisible Unicode tag characters to bypass email security filters. The attackers use these characters to split financial lure words such as 'funding,' preventing email filters from properly parsing and flagging them. This technique repurposes a method typically associated with hiding instructions from humans while exposing them to AI models. The campaign sends millions of emails, making it a large-scale threat. The use of Unicode manipulation represents a novel evasion technique targeting both traditional and AI-based email security solutions. Microsoft's Security Research team identified and reported the campaign. The primary lure appears to be financial in nature, targeting recipients with funding-related content.
A high-volume phishing campaign identified by Microsoft uses 'ASCII Smuggling' — a technique that inserts invisible Unicode Tag block characters (U+E0000 to U+E007F) inside financial keywords (e.g., 'funding' becomes 'fun[U+E0020]ding') to split keyword strings and evade email filters that rely on literal string or regex matching. These characters are invisible to human users and standard rendering pipelines but are processed by email filters and AI/LLM models, causing misclassification. The campaign ran at high volume from approximately early February 2026 through mid-May 2026, following a weekly cadence (active weekdays, silent weekends), generating between 1 and 2.37 million messages per day with a peak on February 26, 2026. Emails were relayed through the ActiveCampaign marketing platform using disposable finance-themed sender domains, with outbound links routed through ActiveCampaign's click-tracking domains (acemlnd[.]com and activehosted[.]com). The campaign targeted Small Business Administration (SBA) loan applicants using AI-generated phishing emails that mimic business loan, line-of-credit, and advance-funding lures, aiming to harvest detailed business and financial information for future spear-phishing. The broader campaign infrastructure leveraged ActiveCampaign's AI-powered marketing automation to mass-produce convincing, tailored phishing websites adapted to different impersonated domains. The technique is adapted from AI prompt injection research, demonstrating that AI-era evasion methods can be weaponized in traditional phishing campaigns. The Unicode Tags block (U+E0000–U+E007F) was originally intended for language tagging and is now deprecated, making it an unusual but effective obfuscation medium.
1. Update email security filters and detection rules to normalize or strip Unicode Tag block characters (U+E0000–U+E007F) before keyword and regex matching, ensuring obfuscated terms are correctly identified. 2. Configure email gateways to flag or quarantine messages containing high densities of invisible Unicode code points as a suspicious signal. 3. Apply additional scrutiny and reputation checks to emails originating from shared marketing automation platforms such as ActiveCampaign, even when they carry valid authentication (SPF/DKIM/DMARC). 4. Block or monitor the identified malicious sender domains and click-tracking domains (acemlnd[.]com, activehosted[.]com) in email and web filtering systems. 5. Educate users to be cautious of unsolicited financial lure emails related to SBA loans, business funding, lines of credit, and advance funding. 6. Implement detection logic that identifies the presence of Unicode Tags block characters (U+E0000–U+E007F) interleaved within email body text as an evasion indicator. 7. Review and harden AI/LLM-assisted email processing pipelines against prompt injection via invisible Unicode characters embedded in email content.
guardiangrowthfunding[.]com, digitalcapitalboost[.]com, thebusinessloanexpress[.]com, yourlocfunding[.]com, advancefundingboost[.]com, guardiancapitalway[.]com, harboradvancefunding[.]com, unitedfundingwave[.]com, directcapitalboost[.]com, onlinedirectfinance[.]com, acemlnd[.]com, activehosted[.]com, Unicode Tags block characters: U+E0000 to U+E007F interleaved within financial keywords
