Security & trust
GovSecure AI is a regulated multi-tenant SaaS for government agencies. This page summarizes the security posture and compliance roadmap for procurement and security reviewers.
Row-level security policies are defined for every tenant-scoped Postgres table, and enforcement on the application connection is being rolled out. Tenant scoping is enforced today in the application layer, keyed to the agency, with defense-in-depth guards and append-only audit logs (ADR-001).
Immutable, hashed audit logs of every action and AI interaction, with role-aware and actor-attributed entries. AuditLog is append-only at the database level.
Role-based authorization (ATO Owner, ISSO, Viewer, Admin, 3PAO) enforced in middleware, server queries, and at the data layer; 3PAO assessors scoped to assigned assessments. WebAuthn multi-factor authentication with recovery codes.
AI outputs are schema-validated and ingested content is delimited against prompt injection; every AI call is audited and usage is metered per-agency entitlement.
TLS in transit; evidence content hashing; secrets validated and server-only. At-rest encryption and secrets management on the InsForge-managed Postgres.
HSTS with preload, a Content-Security-Policy (frame-ancestors none, object-src none), nosniff, and a strict referrer policy on every page.
GovSecure AI does not hold a FedRAMP authorization and is not listed on the FedRAMP Marketplace. The platform is built against NIST 800-53 Rev 5 and OSCAL to help agencies pursue and sustain their own authority to operate.
