Changelog

An updated solution overview deck is now available for enterprise evaluation teams. The deck covers the full Arbitex Gateway architecture: 3-tier DLP pipeline, 8 compliance framework bundles, identity stack, audit chain, and Hybrid Outpost deployment. It is structured for presentations to CISOs, compliance officers, and enterprise architects evaluating AI governance infrastructure. Contact your account team to request the current version.

The Arbitex admin portal now surfaces an account security monitor panel that highlights unusual access patterns across registered users and service accounts. The panel identifies anomalous request volumes, off-hours access from unexpected locations, and service accounts with activity patterns inconsistent with their normal baseline. Security events are captured in the tamper-proof audit log and optionally forwarded to configured SIEM connectors for SOC review.

A dedicated EIN detector is now available in the Arbitex DLP pipeline. The detector identifies nine-digit Employer Identification Numbers across structured and free-text content, including variations with dashes, spaces, and label prefixes. EIN detection supports payroll processing, tax filing, vendor management, and financial reporting workflows where inadvertent EIN exposure in AI requests creates compliance risk under IRS data handling guidelines and financial privacy regulations.

The healthcare DLP bundle now includes expanded detector coverage for specialty clinical identifiers: Clinical Document Architecture (CDA) document identifiers, ICD-10 diagnosis code co-occurrence patterns, lab result value formats, and radiology report structure markers. The expansion improves detection of PHI embedded in structured clinical document formats commonly used in EHR integrations. All 18 HIPAA identifier categories remain covered; the expansion adds detection depth for structured document contexts.

Administrators can now assign risk tier classifications to registered model endpoints in the Arbitex admin portal. Risk tiers (Low, Medium, High, Critical) govern which compliance bundle policies apply, what audit retention periods are enforced, and whether model validation workflows are required before activation. Risk tier assignments are logged with version history and support OCC SR 11-7 and Federal Reserve SR 11-7 model risk management documentation requirements for AI systems in regulated institutions.

A dedicated government industry reference page is now available at arbitex.ai/industries/government. The page covers AI governance for federal agencies and public sector organizations: CUI detection across NIST SP 800-171 category markings, FedRAMP-aligned data sovereignty via Hybrid Outpost, NIST 800-53 Rev. 5 control mapping across AU/AC/SC/SI families, air-gap deployment for classified and IL4/IL5 environments, and tamper-proof audit evidence for FISMA assessments and ATO packages.

The insurance industry reference page at arbitex.ai/industries/insurance covers AI governance for carriers, MGAs, and TPAs across claims, underwriting, and customer service workflows. The page documents GLBA NPI handling for annuity and life products, NAIC Insurance Data Security Model Law alignment, AI governance documentation for underwriting model validation, claims data sovereignty via Hybrid Outpost, and SOX §404 IT controls for insurance holding company structures.

A consolidated compliance frameworks guide is now available in the documentation. The guide maps each supported regulatory framework — HIPAA, PCI-DSS, SOX, GLBA, FedRAMP, FISMA, NIST 800-53, CJIS, and GDPR — to the specific Arbitex enforcement controls, DLP detection tiers, audit log fields, and export formats that apply to each. The guide is structured for compliance officers and solution architects evaluating Arbitex for specific regulatory environments.

A formal model validation workflow is now available in the Arbitex admin portal. Before a new AI model endpoint can be added to routing policy, administrators can require completion of a structured validation checklist: provider risk review, capability documentation, output format verification, and policy bundle compatibility check. Validation records are stored against the model registry entry and surfaced during compliance reviews. The workflow supports OCC SR 11-7 and Federal Reserve SR 11-7 model risk management documentation requirements.

The model registry now supports full inventory export in CSV and JSONL formats. Exports include provider, model version, routing policy assignment, usage volume, validation status, and last-active timestamp for every registered endpoint. Inventory exports support model governance audits, procurement reviews, and compliance filings that require documented AI model inventories. Exports are available from the admin portal and via API.

A structured rollback runbook is now included in the Arbitex deployment documentation. The runbook covers Alembic migration rollback procedures for all supported database configurations, including PostgreSQL on managed cloud services and self-managed deployments. Procedures include pre-migration backup validation, mid-migration abort handling, and post-rollback integrity verification. The runbook is formatted for inclusion in change advisory board (CAB) submissions and disaster recovery documentation packages.

A dedicated insurance industry reference page is now available at arbitex.ai/industries/insurance. The page covers AI governance for carriers, MGAs, and TPAs: GLBA NPI protection for annuity and life products, NAIC Model Law compliance bundle configuration, AI governance documentation for underwriting model validation, and Hybrid Outpost deployment for claims data sovereignty. Compliance mapping covers GLBA, NAIC Model Law, CCPA, NY SHIELD Act, and SOX §404 for holding company structures.

The integrations page has been reorganized into a structured hub with categorized sections covering AI providers, identity protocols, observability platforms, SIEM targets, deployment tooling, and developer SDKs. Each integration entry includes availability status and configuration context. The hub serves as the primary reference for solution architects evaluating Arbitex in their environment. Provider and protocol entries link to relevant feature pages for detailed configuration documentation.

A SIEM integration guide is now available covering audit log delivery to Splunk, Microsoft Sentinel, and Elastic Security. The guide covers webhook configuration for real-time event streaming, field mapping to SIEM-native schemas, alert rule templates for DLP enforcement events, and retention policy alignment. Configuration examples use the Arbitex webhook signing format with RS256 payload verification. Available from the integrations documentation hub.

A model registry is now available in the Arbitex admin portal. The registry catalogs every AI model endpoint in use across your organization — provider, version, routing policy, and usage volume. Administrators can view which models are active, which are being evaluated, and which are restricted by policy. The registry serves as the foundation for model risk governance workflows: each model record can carry documentation links, risk ratings, and approval status. Model registry data is available via the API for integration with external governance tooling.