Security · 3h ago
Old-school text salting tricks still fool AI spam filters
Researchers found that simple text salting techniques, like adding extra spaces or characters, can bypass LLM-powered email filters. These methods, dating back decades, remain effective against modern AI defenses. The findings highlight ongoing vulnerabilities in AI-based spam detection systems.
Meridian48 take
The story underscores that AI spam filters, despite their sophistication, are not immune to basic adversarial tricks—a reminder that security is a cat-and-mouse game.
Read the full reporting
AI spam filters are getting suckered by old-school text salting →
The Register
ai-spam-filterstext-salting