Unicode Tags Fuel Phishing and AI Pipeline Risks

The New Stack reports attackers are inserting invisible Unicode tag characters into phishing emails at massive scale. Microsoft built a prompt injection detector that instead caught one such campaign. The same method can break AI agent pipelines processing untrusted text because the tags remain hidden from standard filters yet readable to targeted systems. The article links the technique directly to real-world email attacks and notes its potential impact on automated workflows. No other details on scale or specific targets appear in the post.