Security & Trust Center
To manage AI as an enterprise and keep using it, you need clear security, access management, and data-handling terms. GSCORA designs each product's control plane around Japanese enterprise security review and continuous AI operations.
GSCORA Content Studio is available now. These are the main security measures in place today.
Authentication
ImplementedAccess is managed through account authentication.
Encryption in transit
ImplementedTraffic is encrypted with TLS.
Data storage
ImplementedStorage of uploaded files and outputs follows the AI Data Policy.
Data deletion
ImplementedYou can delete uploaded files and outputs from your account.
Providers used
ImplementedPer-model data terms are listed in the AI Data Policy.
Contact
ImplementedFor security questions, use the contact form or email sales@gscora.com.
GSCORA API Platform is in Private Beta. The items below are design goals and current validation status — none is presented as fully implemented.
Security Overview
GSCORA API Platform is designed for Japanese enterprise security review. The control plane runs in a Japan region, with encryption in transit and at rest, access control, and audit logs.
Data Flow
Your requests are authenticated, policy-checked, and routed by the GSCORA control plane (Japan region), then sent to the selected model provider. Inference regions differ by model. Control-plane metadata (usage, cost, audit) is processed in the Japan region.
Encryption
Private Beta — Design Partners onlyTraffic is encrypted with TLS 1.2+ and data is encrypted at rest. Credential handling differs by type: GSCORA-issued tokens are stored only as non-reversible verifiers (hashes), never in plaintext. BYOK provider credentials are encrypted at rest with KMS envelope encryption and decrypted only on authorized requests.
Identity & Access
Private Beta — Design Partners onlyRole-based access control (RBAC) manages permissions by department, project, and customer. SSO is planned.
Logging & Retention
Private Beta — Design Partners onlyActions such as API key creation, policy changes, and budget changes are recorded in audit logs. The default retention period is 30 days, configurable from 90 to 365 days per contract. Audit logs contain metadata only — never prompts or generated content.
Incident Response
In developmentWe are formalizing our security incident response process — reporting contact, triage, and notification — and validating it during Private Beta.
Business Continuity
PlannedWe are designing failover to alternate models when a primary model is unavailable, and are putting backup and recovery procedures in place.
Subprocessors
The model providers we use as subprocessors for AI inference are listed in the AI Data Policy.
Data Processing Agreement
Private Beta — Design Partners onlyWe offer a Data Processing Agreement (DPA) to Design Partners, available for review before contract.
Responsible Disclosure
Report vulnerabilities to security@gscora.com. See the Responsible Disclosure page for scope and response targets.
Certification Status
PlannedWe do not currently hold SOC 2, ISO 27001, or similar certifications. SOC 2 control objectives are used as a design reference. Certification is on the roadmap.
For security details or to request a DPA, contact us below.
Contact usData flow by connection mode
Managed API Private Beta
Requests are authenticated and policy-checked in the GSCORA control plane, then sent to the model provider.
BYOK connection Private Beta
Your provider credentials are stored encrypted and used only on authorized requests.
Handling by data type
| Data type | Managed API | BYOK |
|---|---|---|
| Prompt | Passes · not stored | Passes · not stored |
| Completion | Passes · not stored | Passes · not stored |
| Usage Metadata | Stored | Stored |
| Cost Metadata | Stored | Stored |
| Audit Events | Stored | Stored |
Inference-data processing regions and retention vary by model, provider, cloud, and contract. Metadata integration (Observe-only) is a future consideration.