State, local and educational organizations are prime targets for cyberattacks, and phishing — tricking users into revealing usernames, passwords or other credentials — is one of the most common tactics. Using AI, attackers craft messages that are increasingly plausible, combining public information with existing email chains, for example, to impersonate vendors, bosses and colleagues. Worse, the messages often lack the malicious links and attachments that traditionally identified scams.  

AI’s increasingly personalized attacks require a new response. Instead of blocking known hazards only, the solution is to look at communication behavior across a network and flag deviations from normal. Unusual locations, devices or even the type or timing of a message can be suspect. Such an AI-based system can remediate a problem before a suspect email ever reaches its target. And because it is automated, the system can save resources.  

“What we’re seeing more and more now are emails or contacts that contain nothing that is an indicator of compromise. There’s nothing in that email for those tools to look at and say, ‘yes, this is bad,’” said Zach Oxman, Area Vice President, SLED, Abnormal AI. “The only way to stop advanced threats today is to understand behavioral patterns.” 

In this video interview, Oxman discusses the ways AI enables bad actors to tailor targeted, personalized phishing attacks and how an AI-based behavioral approach provides protection. Topics include: 

• How hackers use public information to refine attacks  
• Why traditional security tools are no longer enough 
• How to update cyber defenses to meet new phishing threats