Phishing Campaigns Exploit Text Salting to Bypass AI Email Filters

Since April 2026, over 1 million phishing emails have been detected using a technique called text salting, which confuses AI-powered email security filters. Cybersecurity firm Barracuda Networks reported that attackers are embedding benign-looking words within malicious content to evade detection. This method, while not new, has proven effective against modern LLM-based security systems, which struggle to interpret hidden text. The phishing emails primarily target retail consumers with offers of rewards and gift cards, using compromised domains and standard authentication methods like DKIM. Despite the obvious spam characteristics, these emails are slipping through due to the limitations of current AI detection capabilities. Barracuda suggests that organizations should adopt a layered approach to email security to combat this issue.

Key Points: • Over 1 million phishing emails detected using text salting since April 2026. • Text salting involves hiding benign words to confuse AI email filters. • Organizations are advised to implement layered email security measures.