AI security
Safe AI controls for ChatGPT and GenAI tools
Govern AI adoption without driving users into unmanaged tools or personal devices.
Principle
Do not start by blocking AI entirely. Discover usage, approve sanctioned tools, restrict risky actions such as uploads, and apply prompt-level data protection.
Control Stack
DiscoverUse Shadow IT and Gateway analytics to see which generative AI tools are used.
GovernAllow approved apps while blocking risky actions like file upload or unsanctioned sharing.
Protect promptsApply DLP profiles for credentials, PII, source code, financial data, and customer data.
ExtendUse AI Gateway for programmatic model calls and MCP portals for agent tool governance.
Validation
Confirm sanctioned AI apps work, upload attempts are blocked where intended, sensitive prompts are logged or blocked, and events appear in Gateway/DLP logs.
Analytics Pairing
Use the AI Security Report dashboard as the operating view for this design. It helps confirm which AI applications users actually visit, whether those apps are approved or still waiting for review, how much data is uploaded by app status, and whether MCP servers are protected by Access.