Policies & Rules

Define custom security rules and content policies for your AI applications.

What are Policies?

Policies are customizable rules that define what content is allowed or blocked in your AI interactions. They give you fine-grained control over security and compliance.

Policy Types

Content Policies

Block or filter specific types of content (topics, keywords, sentiment)

Data Policies

Control how sensitive data is handled (PII redaction, data retention)

User Policies

Set permissions and rate limits per user or role

Compliance Policies

Pre-built templates for HIPAA, PCI-DSS, GDPR compliance

Creating a Policy

// Create a custom policy
const policy = await talon.policies.create({
  name: 'block-competitor-mentions',
  description: 'Block any mentions of competitor products',
  rules: [
    {
      type: 'keyword',
      action: 'block',
      keywords: ['CompetitorA', 'CompetitorB'],
      message: 'Competitor mentions are not allowed'
    }
  ]
});

// Apply policy to requests
const result = await talon.analyze({
  content: userInput,
  policies: ['block-competitor-mentions']
});

Policy Actions

Action	Description
block	Reject the request entirely
redact	Remove or mask matching content
flag	Allow but mark for review
log	Allow and log for monitoring

Managing Policies

Policies can be managed via the API or through the TalonAI Dashboard. The dashboard provides a visual editor for creating and testing policies.