Trust
Security
Inwire is designed for private, governed, multi-cloud and hybrid-cloud AI workloads. This page summarizes the safeguards, data handling commitments, and reporting channels we use for the platform.
Last updated: May 7, 2026
Security Contact
Report suspected vulnerabilities, abuse, or security incidents to security@inwire.ai. For privacy questions or data requests, use /privacy-request.
Core Safeguards
- Encryption in transit for service communications where supported by the protocol and deployment.
- Encryption at rest where appropriate for managed storage and customer-configured retained data.
- Role-based permissions, tenant isolation, and access controls for platform resources.
- Audit logging, operational monitoring, and alerting for security-relevant activity.
- Secret management practices for credentials, tokens, API keys, and service-to-service access.
- Vulnerability management, dependency review, secure development practices, and incident response processes.
- Vendor and subprocessor review for services that support platform operations.
AI Inference Data Handling
| Area | Default Position |
|---|---|
| Training on customer content | No training or fine-tuning on Customer Content by default. |
| Prompt and output storage | No persistent storage of standard inference prompts or outputs by default. |
| Operational metadata | Retained for billing, reliability, security, usage analytics, debugging, abuse prevention, and platform operations. |
| Human access | No default human review of Customer Content. Limited access may occur for authorized support, security, abuse, legal, or customer-enabled managed-service workflows. |
| Partner-routed traffic | Processed to provide inference, return responses, report usage, maintain reliability, secure the service, prevent abuse, and support billing. |
| Zero data retention | Where available and enabled, prompts and outputs are processed only to complete the request and are not persistently stored. |
Marketplace and Model Router Readiness
Inwire is preparing provider integrations for model routing marketplaces and gateways, including OpenRouter-style routing workflows. For routed inference traffic, the trust contract is the same: document the endpoint retention mode, honor supported privacy and retention flags, avoid model training on partner-routed prompts and outputs by default, and retain only operational metadata needed for billing, reliability, security, abuse prevention, debugging, and compliance.
Marketplace or gateway partners may exchange request, response, status, usage, routing, and billing metadata with Inwire to deliver the service. Any stronger endpoint-specific commitment, including zero data retention, should be confirmed in the applicable endpoint configuration or written agreement before use.
Customer Responsibilities
Customers are responsible for securing their applications, accounts, API keys, access tokens, endpoints, cloud environments, identity provider configuration, user devices, and submitted data. Customers should configure logging, retention, regions, integrations, and access policies to match their legal and security requirements.
Related Documents
Review our Privacy Policy, Terms, Subprocessors, and DPA pages for additional data protection details.