What's New

Release 2025.12.1 introduces a completely redesigned dashboard experience that puts you in control of you AI governance workflow. This release also completes the Publishing System V4 request lifecycle, enabling end-to-end AI Card review and approval workflows with full audit trails.

My AI Dashboard

The new "My AI Dashboard" replaces the previous static layout with a configurable widget system that adapts to your role and priorities. Users can now customize their dashboard to surface the metrics and workflows most relevant to their responsibilities.

The redesigned dashboard features a modernized visual layout with improved information hierarchy and responsive design optimized for desktop and tablet viewing. Users can now show, hide, and reorder dashboard sections including the:

Quick Start Guide
Discovery Widget
Inventory Widget
Model Testing Widget
Vulnerabilities Widget
AI Card Publishing Widget
Compliance & Trust Frameworks Widget

The new interface guides users through the AI governance workflow in a logical progression: discover AI systems, maintain inventory, test for vulnerabilities, remediate issues, verify compliance, and engage with the community. This structure reduces cognitive load for teams managing large AI portfolios while accelerating decision-making through role-appropriate information prioritization.

All historical data, AI Systems, and metrics remain unchanged and fully accessible through the new interface. This is a front-end redesign only, with no impact on existing APIs or integrations. The new dashboard is the default experience as of this release.

Publishing System V4

Publishing System V4 completes the end-to-end AI Card publishing workflows, transforming Cranium into an authoritative system of record for AI transparency documentation. Organizations can now request, review, approve, and publish AI Cards with complete audit trails and multi-tenant sharing capabilities.

Stakeholders can submit AI Card requests through a standardized workflow that automatically routes to designated reviewers. Reviewers can approve AI Cards for immediate publication, request changes with specific feedback that loops back to the submitter, or decline requests with documented justification. All actions generate comprehensive audit trails that track who made each decision, when they made it, and what rationale they provided.

Request new AI Cards, review submissions, and take approval actions directly within the AI Card interface. The Publishing dashboard displays complete request history with filtering and search capabilities. Audit logs capture every publishing decision and action for regulatory compliance and internal reporting.

Vulnerability Management

The vulnerability assessment workflow now supports closing or ignoring individual vulnerabilities with required attestation reasoning, creating documented evidence of security remediation activities for audit and compliance purposes.

The system tracks who made each close or ignore decisions, when they made it, and what justification they provided. Vulnerability scores automatically recalculate when vulnerabilities are marked as closed or ignored, accurately reflecting your organization's risk posture. Previously closed or ignored vulnerabilities appear in the Resolved Vulnerabilities view, providing a complete audit trail for security reviews.

The Vulnerability Assessment interface has been refined with consistent data loading across Models, Technologies, and Datasets tabs. Resolved vulnerabilities now display correctly on the Technologies tab, and filter states persist properly when navigating between tabs and integrations. Visual consistency improvements enhance readability and reduce confusion when managing large vulnerability datasets.

Bug Fixes & Improvements

This release resolves platform stability issues and includes reliability improvements across VCS integrations and vulnerability management workflows:

Bitbucket Data Center: Fixed authentication and repository retrieval for on-premises Bitbucket deployments
GitHub Integration Improvements: GitHub App installation approvals now process immediately, and organizations names display correctly in Detect AI scan results
Vulnerability Assessment: Resolved vulnerabilities now display correctly on the Technologies tab, vulnerability likelihood scores recalculate immediately when closing or ignoring vulnerabilities, and loading states no longer show incorrect data when switching between BOMs
BOM Management: Fixed deletion operations to properly handle foreign key constraints, and filtering by integration now updates counts correctly
CodeSensor: Improved handling of Jupyter notebooks with large embedded output data to reduce scan timeout failures.

Did this answer your question?