Skip to main content

Mar 17, 2026

Custom AI Models (Early Access)

New at RSA Conference 2026, Custom AI Models let security teams define repeatable, organization-specific email patterns using real examples and natural language—without relying on static rules or changing global detection.

Today, Abnormal’s behavioral AI automatically detects anomalous messages based on communication patterns across your environment.

With the introduction of Custom AI Models, analysts can quickly define organization-specific email patterns using a few example messages and a short description—ensuring similar emails are consistently categorized as Attack, Spam, or Safe.

MKT1415 4 CR Custom AI Models B

What’s New

  • Custom AI models for organization-specific patterns created from example messages and short descriptions

  • Message categorization applying Attack, Spam, or Safe classifications to matching emails

  • Historical message testing validating model performance against recent email activity

  • Flexible model deployment across one or multiple tenants

Custom AI Models are available for Early Access at no cost for Abnormal Inbound Email Security customers today.

See a demo of how it works in action:

Learn more about Inbound Email Security:

Protect Against Evolving Email Threats

See how behavioral AI detects attacks that legacy defenses miss.