P system, combining intent-based detection with agentic capabilities to identify and stop sophisticated attacks across email, collaboration tools and browsers.
CYBER DEFENCE
Proofpoint launches Agentic Collaboration Security to combat AI-era attacks
changes the detection problem. Security needs to understand what an interaction is trying to accomplish, reason over the context around it, and act before the attacker succeeds. Not with disparate tools, but as one system that gets smarter with every decision it makes.”
At the foundation of Agentic Collaboration Security is the Proofpoint Knowledge Graph, which combines threat intelligence covering active campaigns, industry attack patterns and compromised suppliers with organisational context, including business relationships, communication patterns, data access and user risk.
Tom Corn, executive vice president and general manager, Threat Protection Group at Proofpoint
roofpoint has announced its Agentic Collaboration Security
P system, combining intent-based detection with agentic capabilities to identify and stop sophisticated attacks across email, collaboration tools and browsers.
The new system is designed to understand what an interaction is attempting to accomplish, using contextual information to distinguish malicious intent from legitimate activity and take appropriate action.
Proofpoint said the approach creates multiple opportunities to identify attacks that initially appear legitimate. It applies progressively deeper AI reasoning before message delivery and continues protection within the inbox to identify threats that evade traditional defences.
The company said attacks are becoming increasingly difficult to detect because they often resemble legitimate business activity. A compromised supplier, for example, can be used to launch an attack through an existing email thread, while fraudulent payment requests can appear consistent with established business relationships.
The increasing use of AI by attackers is also raising the volume, speed and sophistication of attacks. When an account, conversation and request all appear familiar, Proofpoint argues that behavioural anomalies alone may not be sufficient to identify a threat.
“ Attackers increasingly operate inside the relationships and workflows organisations already trust,” said Tom Corn, executive vice president and general manager, Threat Protection Group at Proofpoint.“ That
The new Nexus Intent-Based Detection Model uses multi-stage analysis to reason over this context, adjusting the depth of analysis according to the ambiguity of each interaction.
According to Proofpoint, most decisions are made in less than half a second, while more ambiguous cases undergo deeper analysis to establish what a message is attempting to accomplish and whether that objective is plausible within its context.
The Knowledge Graph and Nexus model provide a shared foundation through which intelligence gathered at one control point can inform detection, investigation and user protection elsewhere in the system.
Proofpoint said its combination of email gateway and API-based protection allows intent analysis to begin before delivery and continue within the inbox. Intelligence gathered downstream can subsequently strengthen pre-delivery detection.
Intent-based detection is applied at the email gateway before delivery and through APIs within the inbox. For users with significant access or transaction authority, including executives and finance approvers, Privileged User Protection creates dedicated detection models intended to identify highly targeted attacks against specific individuals.
The platform also introduces autonomous threat investigation capabilities. These can reconstruct attacks, identify related messages, determine the potential blast radius and assemble evidence required for response, reducing the manual effort involved in investigating incidents across multiple users.
Adaptive User Protection continuously identifies higher-risk users based on an organisation’ s individual environment, explains the factors contributing to their risk and dynamically adjusts protection and real-time coaching.
32 WWW. INTELLIGENTCISO. COM / MIDDLE-EAST