Shadow AI Monitorfor Chrome
Detects every AI tool your employees use, scans outbound prompts for credential and PII leakage before they leave the browser, and enforces your block/warn/allow policy β all in real time.
Detection is metadata-only. DLP scanning runs entirely in the browser β no prompt content is ever transmitted to AInforce servers.
No Content Capture
Detection metadata only
In-Browser DLP
Scans prompts locally
Policy Enforcement
Block / warn / allow per tool
< 1ms Detection
Sub-millisecond matching
Data leakage prevention
Five-tier DLP classifier, running in the browser
Before a prompt leaves the browser, the extension scans it across five risk tiers. No text is sent to AInforce β detection runs locally using pattern matching on the device.
Policy enforcement
Block, warn, or allow β per tool
Policies are set in the admin dashboard, synced to the extension every 15 minutes, and enforced via Chrome's declarativeNetRequest API for zero-latency blocking.
DNS-level block via declarativeNetRequest. Employee sees a policy violation page.
DLP banner appears before submission. Employee can proceed or cancel with one click.
Tool is approved. Detections are still logged for the risk register.
Default for unclassified tools. Logged and surfaced in the dashboard for review.
150+ AI tools detected
Across every category β chat, coding, agents, MCP infrastructure, meetings, and more. Catalog updates automatically from the AInforce service database.
18 tools
highGenerative AI / Chat
ChatGPT, Claude, Gemini, Copilot, Perplexity
14 tools
criticalAI API / Developer
OpenAI API, Anthropic API, Gemini API, Bedrock
11 tools
highAI Coding
Cursor, Codeium, Devin, v0, Bolt, Lovable
12 tools
criticalAI Agent Platforms
LangSmith, CrewAI, AutoGen, Dify, E2B
Growing
criticalMCP / Tool Execution
MCP Registry, MCP.run and derivatives
9 tools
highAI Productivity
Otter.ai, Fireflies, Fathom, Read AI, tl;dv
8 tools
highAI Writing
Jasper, Copy.ai, Writesonic, QuillBot
13 tools
mediumAI Image / Video
Midjourney, Runway, HeyGen, Synthesia
5 tools
mediumAI Automation
n8n AI, Make, Zapier AI
Installation Guide
Two paths: manual install for testing, MDM deployment for your whole organisation.
Manual Install (Testing)
Download the extension
Open Chrome Extensions
chrome://extensions in Chrome.Enable Developer Mode
Load unpacked
Enter your org key
MDM Deployment (All Devices)
Publish to Chrome Web Store
Google Workspace Admin
Force-install the extension
Push org key via policy
{
"3rdparty": {
"extensions": {
"<EXTENSION_ID>": {
"orgKey": "<YOUR_ORG_KEY>",
"apiEndpoint": "https://ainforce.ai/api/detections"
}
}
}
}Verify in dashboard
What you'll see in the dashboard
Real telemetry from every installed extension flows into your Risk Register and Inventory.
AI Tool Inventory
Every tool detected across your org, with risk classification, detection count, and first/last seen timestamps.
Employee Attribution
When email hash attribution is enabled, see which departments and roles are using which AI tools.
Risk Register
Each detected tool generates a risk register entry with severity, category, and a recommended remediation action.
Preventing Uninstallation
The most common procurement question: "What stops an employee from removing it?" Here's the answer.
Short answer
When deployed via MDM with Chrome's ExtensionInstallForcelist policy, the extension is fully tamper-proof. Chrome prevents users from disabling, removing, or bypassing it. Attempts to visit chrome://extensions show it locked with a managed badge β the Remove button does not appear.
Google Workspace Admin
Chrome Browser Cloud Management
Admin Console β Devices β Chrome β Apps & Extensions
Add extension by Chrome Web Store ID
Set deployment to Force install + Pin to toolbar
Employees see a locked extension β no Remove option
Jamf Pro (macOS)
Google Chrome Managed Preferences
Create a Chrome configuration profile in Jamf
Add ExtensionInstallForcelist payload
Include extension ID and update URL
Push profile to all managed devices
Microsoft Intune
Chrome ADMX Templates
Import Chrome ADMX templates into Intune
Create a device configuration policy
Set ExtensionInstallForcelist with extension ID
Assign policy to device groups
Force-install policy JSON (Chrome registry / plist / GPO)
{
"ExtensionInstallForcelist": [
"<EXTENSION_ID>;https://clients2.google.com/service/update2/crx"
],
"ExtensionSettings": {
"<EXTENSION_ID>": {
"installation_mode": "force_installed",
"update_url": "https://clients2.google.com/service/update2/crx",
"toolbar_pin": "force_pinned",
"runtime_blocked_hosts": [],
"blocked_install_message": "This extension is required by your IT security policy."
}
}
}Replace <EXTENSION_ID> with the 32-character ID assigned when you publish to the Chrome Web Store.
Common questions
What stops an employee from removing the extension?
When deployed via MDM (Google Workspace, Jamf, or Intune) with the ExtensionInstallForcelist policy, Chrome locks the extension entirely. No Remove button appears. Employees cannot disable it, even in developer mode.
Does the extension read what employees type into AI tools?
No. The extension only monitors which domains receive network requests β equivalent to seeing that a website was visited. No page content, no prompt text, no responses are captured or transmitted. Ever.
What data is sent to the AInforce API?
Tool domain (e.g. chat.openai.com), tool name, risk category, and timestamp. Optionally, a one-way SHA-256 hash of the employee email for attribution β the raw email is never sent.
How do I get my Organisation Key?
Log into your AInforce dashboard β Settings β Extension. Your org key is displayed there. Each organisation has one key that all extensions in your org should use.
Does it affect browser performance?
No measurable impact. The request matching is done against an in-memory hash map (sub-millisecond), and telemetry is batched and sent once per minute in the background.
Can we deploy this without employees knowing?
We recommend transparency β tell employees that AI tool usage is monitored for security and compliance purposes. This is required in most jurisdictions (EU, California) and builds more trust than silent monitoring.