AI-SPM

Overview

AI-SPM (AI Security Posture Management) within the AI Center offers CISOs and security teams a holistic view of AI usage within their organization, enabling them to identify risks and enforce security best practices. The dashboard highlights key security metrics, including identified security issues, insights into risky users, and an overall AI Security Posture Score.

Why do you need AI-SPM?

Use the AI-SPM to:

Get insights on high-risk users and activities, helping you prioritize your investigation efforts.
Visualize AI application usage across the organization to maintain compliance and ensure performance integrity.

How does it work?

Here’s an overview of how AI-SPM works, detailing the steps and processes for effective use:

The AI-SPM page displays security issues, user insights, and other relevant data for all previously integrated AI apps.
The AI-SPM dashboards are updated with key security and user data, including the total number of AI applications and any security violations. Additionally, AI-SPM calculates an overall security posture score (ranging from 1 to 100), based on the number of apps monitored by Coralogix and whether they have security evaluations assigned to them.

Accessing the AI-SPM

In the Coralogix UI, navigate to AI Center > AI-SPM.
Use the time picker to select the desired time interval for metrics collection.

Totals

This section includes the following counters:

Total AI Applications – The total number of discovered AI apps (monitored and unmonitored).
Total Security Violations – The total number of security violations recorded across all monitored apps.
AI Security Posture Score – A calculated score based on two factors:
- 50 points if all the user's applications are monitored by Coralogix.
- 50 points if all applications have at least one security policy.
For example, if 1 out of 7 applications lacks a security policy, the user's score will be reduced proportionally, rather than losing the full 50 points.

LLM calls

Visualize your application usage, displaying the total number of LLM calls (prompts and responses) and the call count flagged for security issues. Hover over the graph to see the data distribution for any specific time point.

Total LLM Calls – The total LLM calls (prompts and responses) across all monitored applications.
Prompt Issues – The total number of prompts that contain issues.
Response Issues – The total number of responses that contain issues. For example, if a single prompt contains three different issues, it should still be counted as one issued prompt.

User insights

Gain a clear, concise overview of user data and trends, making it easy to compare app users and identify high-cost, high-spend, and high-risk profiles.

High-Activity Users – Top 5 users who sent the most messages in the application, including the total number of messages sent.
High-Spending Users – The top 5 users with the highest spending (based on the cost of their messages in the application), including their total spend.
Risky Users – The top 5 users with the most security-related issues detected in their messages, including the total number of flagged messages.

Note

This capability is available only if the optional User ID parameter is provided through the Coralogix AI Observability SDK for Python.

LLM Calls

Next Observing E-Commerce Support Chatbot