Model Armor feature availability varies by region due to dependencies on services that could process data outside the region. As a result, some regions have limited feature support. This page provides a breakdown of supported features across all available regions.
For details on which regions Model Armor is available in, see Model Armor locations. To understand the data residency implications, see Data residency and endpoints.
Full support
To use all Model Armor features, create templates in regions with full feature support. The following features are available in fully-supported regions:
- Prompt injection and jailbreak detection
- Responsible AI
- Sensitive Data Protection
- Malicious URL detection
- Child sexual abuse material (CSAM) detection
- Multi-language detection
- Antivirus scanning
Limited support
To ensure strict in-jurisdiction data processing, the following features are not available in regions with limited feature support:
| Feature or detector | Notes |
|---|---|
| Multi-language detection | Prompt injection and jailbreak and RAI detectors support English only. You cannot enable multi-language detection when creating a template in regions with unsupported features. |
| Malicious URL detection | Relies on services that might process data outside the region. You cannot enable malicious URL detection when creating a template in regions with unsupported features. |
| Child sexual abuse material (CSAM) detection | Not available for new templates. Relies on services that might process data outside the region. |
| Antivirus scanning | Files can be sanitized, but without antivirus scanning. |