Securonix launches behavioural analytics to detect human and AI-agent threats
The capability links unusual user and AI-agent activity with identity, data access and asset sensitivity, while keeping consequential response actions subject to human approval.
AEBA is designed to flag instances in which an AI agent uses an unexpected tool, accesses data outside its intended purpose or changes its operating pattern.
NEW DELHI: Securonix has introduced Advanced Behavioral Analytics (ABA), a capability designed to help security teams identify activity that is technically permitted but falls outside normal operating patterns, including actions by enterprise AI agents.
In an announcement issued here on October 12, the company said the capability combines Agent and Entity Behavior Analytics (AEBA) with User and Entity Behavior Analytics (UEBA) to connect signals across systems and time. It is available now.
AEBA is designed to flag instances in which an AI agent uses an unexpected tool, accesses data outside its intended purpose or changes its operating pattern. These signals are assessed alongside identity, permissions, data use, asset sensitivity and threat information.
UEBA examines patterns such as unusual login times, increasing data access and interactions with sensitive assets. The combined approach aims to help analysts investigate compromised accounts, insider risks and abnormal agent activity rather than assess individual alerts in isolation.
Securonix said policies, approvals and audit records remain enforceable throughout detection, investigation and response. Consequential response actions require human approval.
Chief executive Toby Weiss said AI was reducing the time between an attacker gaining access and causing business harm, increasing the need for earlier detection and clear accountability over automated responses.
The platform’s AI security operations analyst, Sam, handles repeatable tasks including alert triage, investigation, evidence collection and case creation. Additional controls coordinate AI-supported workflows and keep actions within defined policies, according to the company.
For authorised insider-risk investigations, its Insider Intent Agent provides supporting context while retaining alternative explanations. The company said individual messages, searches or behavioural changes should be assessed within broader evidence rather than treated as proof of intent.
Securonix also announced ISO/IEC 42001:2023 certification for its AI management system covering its product portfolio, including ThreatQ. The certification followed an external audit, the company said.
Separately, a company survey of 1,000 cybersecurity professionals conducted during summer 2026 found that 96% considered cybersecurity automation important and 49% used AI in behavioural analytics with human approval. The full report and methodology are scheduled for release in November.
