​
AI Security Posture Management

Gain full control of your project’s security with the new AI Security Posture Management (AI-SPM). This feature enables you to monitor and strengthen security across your projects:

  1. Total Security Violations: View the number of security violations in your projects, with clear visual trends showing increases or decreases over time.
  2. AI Security Posture Score: Assess your project’s security with actionable recommendations to boost your score.
  3. Quick Actions Table: Resolve integration gaps, activate missing features, or address security policy gaps effortlessly with one-click solutions.
  4. Security Violations Over Time: Identify trends and pinpoint top security risks to stay ahead.

​
New Policy: Tool Parameter Correctness

Ensure accuracy in tool usage with our latest policy. This policy validates that tool parameters are correctly derived from the context of conversations, improving consistency and reliability in your LLM tools.

​
Dataset Exploration

We’ve enhanced how you manage datasets and added extended features:

  1. CSV Uploads with Labels: Upload CSV files with support for a label column (TRUE/FALSE). Records without labels can be manually tagged in the Exploration tab.
  2. Exploration Tab: Label, review, and manage dataset records in a user-friendly interface.
  3. Add a Session from Session Explorer to Dataset: Click the “Add to Dataset” button in the session details window to add a session from your Session Explorer to an uploaded dataset.

​
Collect Feedback on Policy Findings

Help us improve Guardrails by sharing your insights:

  1. Use the like/dislike button on session messages to provide feedback.
  2. Include additional details, such as policies that should have been triggered or free-text comments.
Directory syncOctober 31st 2024