April 2026
2 min
New Features
- Custom Behavior Controls — Organizations can now create custom behavior checks tailored to their specific policies. Includes an AI-powered behavior builder chatbot, the ability to fork and customize built-in checks, real-time detection badges, and quick-action prompt buttons for faster authoring.
- Hard-Block Policies — A new policy outcome that presents an undismissable overlay, fully blocking the user's action when a critical violation is detected.
- Redirect Remediation — Policies can now redirect users to a designated page when a violation occurs, enabling guided remediation workflows.
- Violation Outcome Tracking — New analytics for tracking violation outcomes (blocked, dismissed, corrected) with a stats endpoint and dashboard filters. Dismiss reasons are now visible in the session activity timeline.
Improvements
- Smarter Prompt Handling — Prompts containing injection patterns are now sanitized rather than fully redacted, preserving more context for policy evaluation.
- Organization Settings Redesign — Overhauled the settings experience with sidebar navigation, a setup checklist, and a consolidated Company Profile page. Refined styles for encryption keys, AI research, and file upload sections.
- Updated PHI Detection — Migrated to improved Protected Health Information detection checks.
- New AI Site Coverage — Added Microsoft Copilot as a recognized AI chatbot application.
- Trusted Sites — Trusted domains can now be added directly from the social engineering dashboard.
- Design Polish — Resolved style regressions and aligned the product UI to updated design standards.
Bug Fixes
- Fixed sidebar navigation causing full page reloads instead of in-app transitions.
- Fixed reference document PDF preview not rendering in the policy editor.
- Fixed violation outcomes incorrectly showing "bypassed" instead of "corrected" after redirect remediation.
- Fixed the behavior controls table re-rendering on every search keystroke.
- Fixed violation queries not scoped by organization, improving multi-tenant data isolation.