Model Access Management

Access Tiers

TierAccess LevelWhoControls
ConsumerQuery the model via API or UIEnd users, applicationsRate limits, input/output filtering
OperatorConfigure system prompts, tools, RAG sourcesApplication developersChange management, review process
AdministratorDeploy models, modify infrastructureML engineers, platform teamMFA, privileged access management
OwnerFine-tune, retrain, access weightsML research teamHighest privilege, audit everything

Principle of Least Privilege for AI

  • Users should only access AI capabilities required for their role
  • Models should only access data required for their function
  • Tools should be scoped to minimum necessary permissions
  • System prompts should be modifiable only through change management

Model Weight Security

Model weights are the most valuable AI asset. Treat them like source code:

  • Store in encrypted, access-controlled repositories
  • Track all access with audit logs
  • Use signed model artifacts to detect tampering
  • Separate development, staging, and production model stores
  • Implement break-glass procedures for emergency weight access