Skip to main content

    Use AI Without Exposing Confidential Data.

    Locke detects and masks sensitive information before it reaches an AI — 100% on your device, with no data ever leaving your machine.

    Get early access — Locke
    See how it works

    The Problem

    You've already shared more than you think

    Sensitive data flows into AI tools every day—without oversight, controls, or visibility. Here's what that looks like.

    Unapproved AI Usage
    47%

    of genAI users rely on personal, unsanctioned AI apps

    223/mo

    average sensitive-data-to-AI-app incidents per organization — double the rate a year ago

    Most AI adoption occurs without formal IT approval or oversight — and incidents are accelerating.

    What Gets Shared
    34.8%

    of corporate data employees paste into AI tools is classified sensitive

    27.4%

    the same figure just one year earlier

    Source code is the single largest category of sensitive data employees paste into AI tools.

    The Cost of Exposure
    $670K

    additional cost per AI-related data incident

    CVSS 9.3 / 9.6

    zero-click prompt-injection exfiltration and RCE, already found in production AI agents — this is no longer hypothetical

    AI-related data incidents carry measurable financial, regulatory, and now exploit-grade technical consequences.

    Zero-Knowledge

    We never see, store, or log your data

    100% Local

    Nothing leaves your device

    AI Compliance

    Visibility and control over AI data exposure

    GDPR Ready · HIPAA Compatible

    The solution

    A privacy layer for AI

    Locke sits between you and every AI tool you use. Before any prompt or attachment reaches an AI, Locke:

    Detects sensitive data in real time

    READYLocke status indicator: Protected - sensitive data masked
    Waiting for input...
    Type a message...
    REAL-TIMEON-DEVICE

    Masks sensitive data so the AI still understands the request

    Cloak Active
    SSN
    123-45-6789{{SSN_CLIENT-BILL_1}}
    Email
    john@email.com{{EMAIL_CLIENT-BILL_1}}
    Phone
    (555) 123-4567{{PHONE_OPP-COUNSEL-DISC_1}}

    Detecting sensitive data...

    Prevents third-party scripts from capturing typed content

    Script Guard Active
    addEventListener('keydown')Blocked
    clipboard.readText()Blocked
    input.value (polling)Blocked
    onPaste overrideBlocked

    One local layer for every AI app (Locke — coming soon)

    Locke — coming soon
    Your AI apps
    Locke
    AI provider
    Works across every AI app on your machine
    Detects and masks before anything is sent
    No external servers — 100% on-device

    The device boundary

    Your data never crosses this line.

    Locke detects and cloaks sensitive data before it leaves your device — only the cloaked version ever reaches an AI provider, and only your device can restore the real reply.

    Diagram: Your AI apps send data to Locke, running on your device. Locke detects sensitive data and cloaks it with format-preserving encryption before it crosses the device boundary to the AI provider. The AI provider's reply crosses back into the device still cloaked, where Locke restores the real values before handing the reply back to your AI apps. Raw sensitive data never leaves the device boundary.

    Detect (multi-engine) · Cloak (format-preserving encryption) · Restore (in the reply)

    Try it yourself

    See it work. Right here, right now.

    This demo runs entirely in your browser — open devtools, watch the network tab: nothing leaves this page.

    What you'd type
    Try:
    What the AI would seeProtected
    Draft a demand letter for my client Ms. Zquce Bysktikw (SSN 310-57-1596), who lives at 128 Cluhzcvbh Terrace and can be reached at tmmuf.nagmthsr@gmail.com or (067) 478-9216. Her medical bills total $42,882 for Case No. 4457-7981.
    Person nameSocial Security NumberStreet addressEmail addressPhone numberDollar amountCase number

    7 detections · 7 types · Network requests made by this demo: 0

    ReversibleLocke restores real values once the AI's reply comes back

    Sure — here's a draft referencing Ms. Zquce Bysktikw. Let me know if you'd like any changes.

    Simulation of Locke's on-device engine. The real product uses FF1 format-preserving encryption (NIST SP 800-38G) and restores real values in the AI's reply.

    Works with the AI tools you already use · Locke (Desktop App)

    Works everywhere you use AI

    Locke, the Sonomos desktop app, runs locally between you and the AI tools you already use — desktop apps, browser sites, IDEs, and terminals. No integrations, no setup, nothing leaves your device.

    Locke is compatible with

    Desktop apps

    ChatGPT
    Google Gemini
    Claude
    Microsoft Copilot
    Perplexity
    more

    Browser websites

    ChatGPT
    Claude
    Google Gemini
    Grok
    Perplexity
    Microsoft Copilot
    more

    AI-native IDEs & code editors

    VS Code
    Cursor
    GitHub Copilot
    more

    Traditional IDEs

    running Claude Code & Codex

    Visual Studio
    PowerShell ISE
    more

    Terminal consoles

    running Claude Code & Codex

    Windows Terminal
    Git Bash
    Bash
    more

    Locke covers web-based AI platforms, native desktop apps, AI-native IDEs, traditional IDEs and terminal consoles running Claude Code and Codex, and other AI tools on your machine — all processed 100% locally.

    See Locke forChatGPTClaudeGeminiGitHub CopilotCursorall tools

    PARTNERED WITH LEADING ORGANIZATIONS

    Clio logoJD Supra logoMikeCheck logoWalkabout logoIx logo

    Get Started

    Be first through the door.

    Locke, our desktop app, brings system-wide AI privacy to your whole machine — coming soon. Request early access to lock in launch pricing. Curious how much PII you leak while coding? Try Canary, our free experimental Claude Code plugin.

    Get early access — LockeTry Canary — our Claude Code plugin