Overview
ManagedMethods Content Filter monitors and presents details on use of large language models(LLM). This LLM detection can help admins evaluate or investigate inappropriate AI usage.
Detail
Certain LLM inputs or volume of usage may result in admin concerns. AI and LLMs are becoming a popular resource and hub for user activity.
LLM activity is presented under the "Account Details" and "Violations" sections.
Data
Selecting the "AI" subtab will expose a summary of results related to LLM usage.
This section provides a granular look at prompts submitted between the user and LLM service.
Column | Description |
Date | Timestamp of the recorded AI interaction. |
Url | The specific AI platform domain accessed. (e.g., |
Input | The exact text or "prompt" submitted by user to the AI. |
Risks | Indicates if any predefined Policy scan keywords were detected within the prompt. |
External Ip | The public-facing IP address of the network the user was connected to during the session. |
Device | The unique hardware identifier of the device used to access the AI service. |
Actions | Click View to pin the detailed data point for future review. |
Report
Summary reviews may expose educational dishonesty or critical safety and security concerns.
Data results can be exported using the "Reports" option. Files are generated as CSV or PDF format. Reports can also be Scheduled.
Notes
The following LLM sources are currently monitored.
Chatgpt, Claude, Gemini, Copilot