Proofpoint Announces Agentic Data, AI, and Collaboration Security

Proofpoint announced agentic data-and-AI and collaboration-security systems using knowledge graphs, intent detection, and human-approved controls.

On 22 September 2026 Proofpoint announced two related but distinct agentic offerings rather than a single product. The Agentic Data and AI Security system connects AI activity, data sensitivity, identity, access, and intent in a shared knowledge graph; detection, investigation, and remediation agents prioritize risky actions, reconstruct events, and recommend access or DLP changes with human approval, and Semantic Business Policies convert plain-language rules into runtime controls. Proofpoint said 87% of organizations have moved AI assistants beyond pilot, while 52% are not confident their controls could detect a compromise. Separately, Agentic Collaboration Security, planned as a Q1 2027 update, pairs a Knowledge Graph with the Nexus intent model to judge whether an interaction is plausible before email delivery and in the inbox, including dedicated models for executives and finance approvers. That release also describes autonomous investigation, adaptive user coaching, red-team and blue-team agents, and Advanced Browser protection with Push Security for post-click phishing, malicious extensions, OAuth phishing, and session hijacking. The two reports do not contradict each other; they describe different systems.

• On 2026-09-22 Proofpoint announced the Agentic Data and AI Security system, linking AI activity, data sensitivity, identity, access, and intent in one knowledge graph.
• Detection, investigation, and remediation agents prioritize risky actions, reconstruct events, and recommend access or DLP changes only with human approval.
• Semantic Business Policies turn plain-language rules into runtime AI controls.
• Proofpoint said 87% of organizations have moved AI assistants beyond pilot, while 52% are not confident their controls could detect a compromise.
• Separately, Agentic Collaboration Security is planned as a Q1 2027 update combining a Knowledge Graph with the Nexus intent-based detection model.
• Intent analysis is meant to run before email delivery and continue in the inbox, with Privileged User Protection models for executives and finance approvers.
• The collaboration release also describes autonomous investigation, adaptive user coaching, and red-team and blue-team agents.
• Advanced Browser protection, built with Push Security, is pitched to cover post-click phishing, malicious extensions, OAuth phishing, and session hijacking.