iDox.ai GuardrailFree for a limited time · 14-day full access

Stop AI data leaks before they happen.

Guardrail is a lightweight app that runs quietly in the background and shields everything you send to AI. It keeps your sensitive data out of ChatGPT — and watches what AI agents do under the hood.

Runs on-device · No credit card · Windows, macOS & Android

Built for compliance
GDPRHIPAAFOIASOC 2*ISO 27001*More →
The shadow AI problem

Your team is pasting secrets into ChatGPT

People move fast with AI — and confidential data goes with it. Once it's pasted or uploaded, you can't take it back.

11%

of everything employees paste into ChatGPT is confidential company data.

Source: Cyberhaven
77%

of AI users paste data in — and 22% of those pastes contain PII or payment data.

Source: LayerX
8.5%

of all AI prompts contain sensitive data, often without anyone realizing.

Source: Harmonic
It already happened.

In 2023, Samsung engineers pasted confidential source code into ChatGPT — and the company banned generative AI overnight. One paste is all it takes.

What Guardrail does

Three layers of AI protection, one quiet app

Real-time AI data loss prevention that goes beyond the chat box — it also watches the AI agents working behind it.

Block data leaks

Catches sensitive text and files — names, SSNs, financials, source code — and masks or blocks them before they reach ChatGPT.

PII redaction · DLP for ChatGPT

Govern AI intent

Detects prompt injection and unsafe instructions hidden in text, code, documents, or commands — and blocks the action before an AI can carry it out.

Prompt injection prevention · AI intent governance

Control AI agents

Watches the hidden chatter between AI agents, and blocks unauthorized commands or privilege escalation.

AI agent security
Prompt injection prevention

AI intent governance: it's not just what data leaves — it's what AI is told to do

Guardrail reads the meaning behind a prompt, not just its keywords. It examines 10 categories of high-risk intent — across plain text, code, documents, and commands — and stops them before an AI agent can act.

Payments or purchases
Installing software
Establishing remote connections
Modifying system settings
Uploading sensitive documents
Logging into an account
Creating accounts
Deleting local files
Modifying permissions
Sharing personal data
Detected in plain textDetected in codeDetected in documentsDetected in system commands
Editions

Pick the right fit

Start free as an individual, or roll out central policy and audit across your whole company.

★ Most popular

Individual

Real-time Guardrail for pros & small teams
  • On-device interception of sensitive data
  • Real-time auditing of AI agent activity
  • Abnormal command & prompt-injection detection
  • Windows, macOS & Android
Explore Individual
For teams & orgs

Enterprise

Centralized governance for enterprise AI
  • Management console — policy engine, roles, audit
  • Role-based access per team, full Who-What-When trail
  • Hosting service (cloud) or on-premises deployment
  • Vertical policy packs for GDPR, HIPAA, FOIA & more
Explore Enterprise
Get started in 60 seconds

Download Guardrail — free for a limited time

Pick your device and install. No credit card, no setup headache — protection starts the moment it's running.

macOS

Apple Silicon Macs only (M1 and later)

Download for macOSTrouble installing? Get help →

Android

Get real-time alerts on the go

Get it on Google PlayTrouble installing? Get help →

Prefer the command line?

Power users on Windows can install with a single PowerShell command.

iwr -useb https://idox.ai/install-guardrail.ps1 | iex
FAQ

Questions, answered

Stop AI data leaks before they happen

Join the individuals and enterprises keeping their data safe while they move fast with AI. Free for a limited time.