Overview
Shield Sentinel is the student-safety monitoring component of Shield. It watches for signs of serious concern — things like self-harm language, bullying, grooming, or violent threats — inside a school's Google Workspace, and routes alerts to the counselors, administrators, and parents who need to know.
What Sentinel Monitors
Sentinel connects to your school's Google Workspace through a read-only integration authorized by your IT partner (the MSP). Once connected, it periodically checks:
- Gmail — School email messages sent and received from student accounts
- Google Drive — Documents created or shared in the school's Workspace
- Google Chat — Messages exchanged in school Chat spaces
Every piece of content passes through an AI analysis pipeline. The vast majority of content produces no signal at all (classified as noise and discarded). Only content that clears a series of increasingly rigorous checks results in an alert.
What Sentinel Does NOT Monitor
Sentinel has a strictly defined boundary. It does not look at:
- Personal accounts — Personal Gmail, personal Google Drive, personal YouTube, or any non-school account. Sentinel only has access to accounts within the school's Google Workspace domain.
- Device content — Files stored locally on a student's device, photos, text messages, social media apps, or anything else outside the school's Workspace. Sentinel is not device surveillance software.
- Microsoft 365 or other platforms — Currently only Google Workspace is supported. Microsoft 365 support is on the roadmap.
- Off-campus personal activity — If a student uses a personal phone or personal computer on a home network, that activity is completely outside Sentinel's scope.
The Four Alert Tiers
Not every flagged signal becomes an alert. Sentinel uses a four-tier system to separate genuine concerns from noise:
| Tier | Name | What it means |
|---|---|---|
| Tier 0 | Noise | No alert. The AI determined there is no genuine concern, even if a keyword matched. |
| Tier 1 | Monitor | A mild or ambiguous signal. Logged for trend-watching and shown in the Sentinel alert inbox. |
| Tier 2 | Alert | A counselor-level concern. It appears in the Sentinel alert inbox for the counseling team to review. |
| Tier 3 | Critical | Immediate concern such as self-harm, grooming, or imminent violence. It appears in the Sentinel alert inbox for counselors and administrators to review. Parents linked to the student can also see it in the parent portal. |
The tier reflects both how serious the AI judged the content and what kind of action it recommends.
How Student Privacy Is Protected
Sentinel is designed so that no raw student content ever persists to a database. Messages and documents are fetched into memory, analyzed, and discarded. What gets stored is analysis metadata — categories detected, severity, timestamps, and a one-way hash of the student's email address. Counselors see AI-assembled case summaries, not original messages.
Frequently Asked Questions
Does Sentinel read every single email a student sends?
Sentinel scans content from the school's Google Workspace accounts. Most content produces no signal. Only content that triggers a keyword match moves on to further analysis, and most of that is filtered out as noise before an alert is ever created.
Can parents use Sentinel to read their child's messages?
No. Parents see a summary of alert categories and any notes the counseling team shares. Raw message content is never shown to anyone, including parents.
What if a student uses a VPN or a personal device?
If the student is logged into their school Google account, Sentinel can still see activity in that account. Activity on personal accounts or through personal apps is outside Sentinel's reach.
Who authorized the connection to our Google Workspace?
The school's Google Workspace administrator approved the OAuth connection during setup. The integration uses read-only access scopes.