Overview
This latest update to AI Access extends AI Guardrails beyond prompts and responses to include files uploaded to Generative AI applications.
Packaging and Licensing
This update is included as part of the Secure Access Advantage package.
Key Benefits
This enhancement strengthens data protection by addressing file uploads, which are often a source of data exfiltration and compliance risks. It ensures that sensitive information within uploaded files is subject to the same robust Security, Privacy, and Safety rules already applied to prompts and responses.
Use Cases
- Prevent the upload of files containing sensitive data (e.g., PII, confidential documents) to unauthorised AI applications.
- Monitor file uploads to understand how users are interacting with AI services and identify potential policy violations.
Getting Started
With this update, all AI Guardrails rules configured to scan prompts will now automatically scan uploaded files as well. No additional configuration or changes to existing AI Guardrails rules are required.
Documentation and Resources
Best Practices
- Start by configuring rules in Monitor mode to understand the types of file uploads occurring in your environment and the potential policy violations.
- Leverage the AI Guardrails Data Classifications effectively to target specific types of sensitive data within files.
- Configure user notifications to educate users about acceptable use policies for AI applications.
- Regularly review monitored events and refine your rules based on observed activity.
