Guardful AI

Scan policies

Choose a risk check template, or create your own custom set of checks

Templates

Start with a template, then modify the checks if needed

TemplateChecks
Basic Data ProtectionEmail addresses, Phone numbers, IP addresses, Payment card numbers, IBANs, …
Developer Log GuardEmail addresses, IP addresses, Payment card numbers, API token formats, Private key markers, …
Prompt Attack ReviewPrompt attack risk
Balanced AI InputEmail addresses, Person names, Prompt attack risk, Toxic language, Threats
Support Output ReviewEmail addresses, Phone numbers, Toxic language, Insults, Violence
Refund approval reviewStructured business rules, Custom text policies
Escalation reviewCustom text policies
Delivery promise reviewCustom text policies
Supported product claimsCustom text policies
Confidentiality reviewCustom text policies
Consent and disclaimer reviewStructured business rules, Custom text policies
Assistant topic boundariesCustom text policies
Community policy exceptionsCustom text policies

Custom scan policy

Choose any checks and which actions to take for them

Illustration: AI attacks is checked and assigned Block; Harmful content is checked and assigned Review; Sensitive data is unchecked