Default Security Groups and Rules
AI Model Security uses Security Groups with one-to-one relationships to Model Sources
(Local, Hugging Face, Azure, GCS, S3) to establish and configure security controls for
models.
| Where Can I Use This? | What Do I Need? |
|---|
- Prisma AIRS (Model Security)
|
- Prisma AIRS Model Security License
|
AI Model Security leverages Security Groups to enable you to establish and configure your
Model Security controls. Security Groups have a one-to-one relationship with Model
Sources and this association is permanent once set.
Each Model Source provides specialized metadata that is used in rules and security
validation processes.
AI Model Security supports one default Security Group per source and these security
groups are already created for you.
AI Model Security includes a default Security Group for every source type. You can
create additional custom security groups through the Strata Cloud Manager user
interface, the Python SDK, or the CLI.
- (Default) Local Storage
- Hugging Face
- Amazon S3
- Google Cloud Storage
- Azure Blob Storage
- JFrog Artifactory
- GitLab Model Registry
Your Strata Cloud Manager tenant includes the following pre-configured default security
groups.
| Security Group Name | Model Source |
|---|
| Default LOCAL | Local Storage |
| Default HUGGING_FACE | Hugging Face |
| Default S3 | Amazon S3 |
| Default GCS | Google Cloud Storage |
| Default AZURE | Azure Blob Storage |
| Default
ARTIFACTORY
| JFrog
Artifactory |
| Default
GITLAB | GitLab
Model
Registry |
