AI Security
How Guard's AI assistant keeps conversations in your tenant, limits each agent's tools, and logs every message.
- AI Data Privacy and Tenant IsolationHow Guard stores AI conversations in your tenant's partition and anonymizes findings before the AI knowledge system uses them.
- AI Operational GuardrailsHow Guard limits each AI agent's tools, depth, run time, and output size, and blocks scans of protected domains.
- AI Output Accuracy and GroundingHow Guard's AI assistant grounds answers in your platform data and retries structured output that fails validation.
- AI Audit Trail and ObservabilityHow Guard records every AI conversation, links sub-agent work to its parent, and ties each trace to the requesting user.
- Responsible AI PrinciplesGuard's AI principles: you confirm active scans, each agent gets only the tools it needs, and conversations are stored per tenant.
- AI Model Providers and Data HandlingWhich providers run Guard's AI models, what each one does, and the contract and service terms that keep them from training on your data.