Abnormal AI launches autonomous agents to advance email security

Abnormal AI has announced new autonomous AI agents designed to transform employee training, risk reporting, and email security for organisations.

The company has launched two AI-powered tools, AI Phishing Coach and AI Data Analyst, to address pressing security challenges posed by the surge in malicious AI-driven cybercrimes over the past year.

Evan Reiser, Chief Executive Officer and Founder of Abnormal AI, said, "The most dangerous attacks don't target firewalls - they target people, and people need better protection. Today, we're introducing true AI agents that not only protect people from advanced cyberattacks, but also eliminate the manual effort that's bogging down security teams. From personalised phishing simulations to autonomous reporting and expanded threat remediation capabilities, these innovations represent a massive leap forward in how AI can be operationalised across the security stack."

According to a recent survey, 53% of security leaders believe the resources required for existing security awareness training do not correspond with their impact. Abnormal AI's new AI Phishing Coach aims to address this by providing a personalised approach. It creates simulation-based training scenarios derived from actual attacks the Abnormal platform has blocked and delivers immediate coaching modules when users interact with suspicious content, thus eliminating the reliance on standardised training videos or generalised courses.

For broader organisational education, the platform offers AI-generated, branded video training tailored to each company's threat profile. Unlike traditional training systems that depend on dated templates, the AI Phishing Coach uses real-time behavioural threat intelligence to adapt training content and timing, functioning as a tailored security mentor for each employee.

Abnormal AI also introduced AI Data Analyst, a tool designed to distil complex security data into accessible and actionable insights. The AI Data Analyst provides proactive report generation, allowing customers to receive tailored impact assessments. It also supports interactive exploration, enabling users to pose specific queries or to request presentation-ready reports assembled with clear language and interactive features, illustrating Abnormal AI's contribution to organisational security.

Another addition to the platform is a set of enhancements aimed at strengthening email security. The company's Behaviour Platform has reportedly outperformed legacy secure email gateways (SEGs) and even human analysts, prompting three-quarters of its customers to transition away from third-party SEGs.

To enhance this transition, Abnormal AI has announced three updates, available at no additional cost. The Quarantine Release feature consolidates Microsoft-quarantined emails into the Abnormal platform for simplified incident triage and response. The new URL Rewriting capability introduces user-facing alerts and click tracking for potentially risky links while preserving user experience. Enterprise Remediation Settings enable administrators to customise remediation actions based on the nature of the threat and organisational requirements.

In the area of international expansion, the company has recently achieved FedRAMP Moderate Authorisation in 256 days, facilitating adoption by US federal agencies. The company is also expanding its operations in Germany, with planned entries into Japan and France later this year. Localisation of the Behaviour Platform for languages and regional threat profiles is underway.

Abnormal AI reports its platform is currently in use with over 3,200 organisations worldwide, including more than 20% of Fortune 500 companies. The company's anomaly detection technology evaluates identity and behavioural context to assess the risk of every cloud email event and prevent complex socially engineered attacks.

The platform is deployable via API integration for Microsoft 365 and Google Workspace, with additional coverage for services including Slack, Workday, ServiceNow, and Zoom. Abnormal AI aims to provide security teams with new levels of automation in workforce training, executive reporting, and email protection against advanced threats.