AI Security Use Cases¶
Useful Links
- For onboarding refer to the AI Security Onboarding Guide
- For list of supported platforms refer to the AI Security Support Matrix
AI Identity Security and AI Compliance & Governance (AI-GRC) are on the roadmap and not in the console today. Plan coverage around the six above.
Which Module for Which Job¶
| What you need to do | Module | You are covered when |
|---|---|---|
| Check a public model before adopting it | ML Model Static Scans | Only scanned and approved models reach your registry |
| Make that check impossible to skip | Model scan in CI/CD | A model enters the registry only through a merged pull request |
| Test a model against jailbreaks and injection | AI Red Teaming | High-risk attack paths are found and fixed before production |
| Stop harmful prompts and data leakage in the request path | Prompt Firewall | Unsafe prompts and responses never reach the user or the model |
| Constrain what an autonomous agent may execute or reach | Agentic AI Security | Agents act only inside the process, file, network and domain limits you set |
| Detect abuse of AI services from cloud events | AI-DR | Runtime threats are detected and routed to a named owner |
| Find AI tools nobody approved | Shadow AI Discovery | Unapproved AI usage is visible in the inventory |
| Keep an inventory of models, agents and datasets | AI-SPM | Every AI asset has an owner and a risk classification |
Not sure which module reaches your assets?
The AI security architecture page maps the four collection methods to the four kinds of asset, which is what decides whether you install an agent, a browser extension, or nothing at all.
