1M+ Emails Use Hidden Text to Dupe AI Security Filters

Barracuda researchers have observed a large retail-themed phishing campaign (over 1 million emails) that evades secure email gateways and AI content detectors by "text salting"—inserting hidden, benign-looking content via HTML/CSS tricks (zero-size fonts, off-screen overflow, layered techniques) to dilute spammy language; attackers are using LLMs to scale and automate these obfuscation techniques, and defenders are advised to assess full message context (visible vs hidden content, links, sender behavior) rather than rely on simple keyword scans.