Overview
Sample Foundation · last 30 days · monitor and redact sensitive prompts across sanctioned and shadow AI tools.
Microsoft 365 E5
Copilot Premium
ChatGPT Business
Intune
SentinelOne
Active users with agent
78 / 80
97.5% of fleet
Sensitive prompts redacted
412
Last 30 days
High-severity events needing review
2
Click to open Activity Log
AI tool usage (last 30 days)
Share of total prompts. Shadow tools have no enterprise SSO.
Microsoft Copilot Premium— 64%
ChatGPT Business— 22%
Claude— 8%
Shadow AI
Gemini— 4%
Shadow AI
Perplexity— 2%
Shadow AI
Usage by department
Total prompts last 30 days
Recent redaction events
Names tagged [DEMO] are seeded for this preview.
| When | User | Department | Tool | Detail | Action |
|---|---|---|---|---|---|
| May 5 | L. Park [DEMO] | Grants Management | ChatGPT Business | Pasted grantee SSN + EIN drafting MOU | Redacted |
| May 5 | M. Hayes [DEMO] | Program Officer — Housing | Microsoft Copilot Premium | Beneficiary home address + medical condition in grant memo | Redacted |
| May 4 | R. Cohen [DEMO] | Communications | Microsoft Copilot Premium | Donor list pasted to draft thank-you newsletter | Anonymised |
| May 3 | A. Weiss [DEMO] | Executive Office | Perplexity | Grant decision summary w/ health condition | Redacted |
| May 2 | T. Mendel [DEMO] | Legal & Compliance | Microsoft Copilot Premium | 990-PF draft w/ grantee finances | Redacted |
Complements Microsoft Purview (data classification) and SentinelOne (endpoint). Prompt Shields handles the in-prompt PII layer those don't see.