Description of the Service
Information Technology Services (ITS) provides access to Azure AI Foundry to members of the University community to support instructional, research, and other University-related activities. This service enables authorized departments, colleges, and units to access Microsoft Azure AI Foundry through a dedicated Azure Subscription, empowering them to provision, manage, and operate their own AI models.
Azure AI Foundry is a managed platform-as-a-service that provides access to a curated catalog of AI models, including models from OpenAI, Meta, Mistral, and Anthropic (e.g., Claude), as well as tools for fine-tuning, prompt engineering, and deploying AI-powered applications. ITS provisions the underlying Azure Subscription and baseline AI Foundry configuration. Departments are then responsible for managing their model deployments, usage, and associated costs within their provisioned environment.
Requests for Azure AI Foundry access require an Azure Subscription for the requesting department. ITS will provision the Subscription and configure the Azure AI Foundry hub and project structure in alignment with University security standards, enabling departments to begin deploying models immediately upon approval.
Key Features
- Provisioning of dedicated Azure Subscriptions scoped to Azure AI Foundry for University use
- Access to the Azure AI Foundry model catalog, including OpenAI, Meta Llama, Anthropic Claude, and other foundation models
- Department, college, and unit level self-service management of AI model deployments via the AI Foundry portal or API
- Centralized hub provisioning and baseline configuration by ITS
- Access to GMU DNS, Identity Management, and other core services
How Do I Get This Service?
To request this service, submit an Azure AI Foundry request through the ITS Service Catalog. Requests may include:
- A new Azure Subscription scoped for AI Foundry use
- Azure AI Foundry hub and project provisioning
- Access to specific models within the AI Foundry model catalog
- Any combination of the above
Requests are subject to review by Cybersecurity Operations to ensure alignment with University security requirements and must also receive approval from the requesting department’s designated business approver prior to provisioning.
Policy
- All costs associated with Azure AI Foundry usage, including tokens, required logging and security controls, compute, and storage, are billed directly to the requesting department or project.
- Requestors are responsible for monitoring usage, managing consumption, and ensuring appropriate funding is available for ongoing AI services.
- Customers are responsible for workload design, data classification, model selection, prompt design, output review, and the operational management of AI applications deployed within their Azure AI Foundry environment.
- Use of AI models through this service must comply with University IT policies, information security standards, and applicable data governance requirements. Departments are responsible for ensuring that data submitted to AI models is appropriately classified and permitted for use with cloud-hosted AI services.
- Azure AI Foundry resources must comply with University IT policies and information security standards.
- University Policy Number 1114: Data Stewardship
Additional Information
AI Foundry Knowledge Base: Azure AI Foundry Resources - Information Technology Services
Customers are encouraged to evaluate AI models and understand the costs associated with token consumption.
Shared Responsibility Model: What Microsoft, ITS, and you are responsible for Azure AI Foundry — Shared Responsibility and Governance Model - Information Technology Services
Microsoft Model and Token Cost Summaries: Foundry Models | Microsoft Azure
Requestors will require an owner of the service who can manage access, financial operations, and other administrative functions.
ITS provides subscription provisioning and baseline Azure AI Foundry configuration but does not manage day-to-day AI application operations, model deployments, or customer-developed AI solutions within the provisioned environment.
Departments unsure of which AI models are appropriate for their use case are encouraged to consult with ITS prior to submission to identify the right model tier and deployment configuration for their needs.
*ITS Core Services, such as DNS and identity management, depend on the data being stored and the use of the service. Research with compliance requirements, such as CMMC 1 or higher, or workloads involving sensitive or regulated data, will likely not have access to many of these central ITS services and should contact ITS to discuss appropriate configurations before submitting a request.