Skip to main content

    Privacy everywhere you use AI.

    Locke 1.0 is out.

    It finds the sensitive data in what you send to the AI tools it covers, and masks it on your device before anything leaves.

    Free for one person. Shipped November 12, 2026, for 64-bit Windows and Linux. macOS is next: get notified.

    See the green glow on the edge as it sends? That's Greenlight, Locke's light on the side of your screen. It flashes the moment something is masked, so you know Locke's on it and you can keep going.

    Partnered with

    Partners, not customers or endorsements. About our partners

    Clio logoJD Supra logoMikeCheck logoWalkabout logoIx logo

    You've already shared more than you think

    Sensitive data goes into AI tools every day. Almost nobody checks.

    1. 47%

      Unapproved AI usage

      of genAI users at work use personal AI apps their employer does not manage

      The employer has no say over what goes into those apps, or how long it stays there.

    2. 39.7%

      What gets shared

      of employee interactions with AI tools involve sensitive data, counting prompts, pastes and file uploads

      The average employee enters proprietary information into an AI tool about once every three days.

    3. $670K

      The cost of exposure

      higher average breach cost at organizations with high levels of shadow AI than at those with little or none

    Masked before the request leaves

    On supported routes, Locke picks the AI requests out of your traffic and masks the sensitive values in them.

    The values go. The meaning stays.

    Each value becomes a one-way placeholder that keeps its meaning. The AI never sees the real one.

    Masking active
    SSN
    123-45-6789{{SSN_CLIENT-BILL_1}}
    Email
    [email protected]{{EMAIL_CLIENT-BILL_1}}
    Phone
    (555) 123-4567{{PHONE_OPP-COUNSEL-DISC_1}}

    Detecting sensitive data…

    Raw data never crosses this line

    Detection and masking run on your device. Only the masked version reaches the AI provider, and its reply comes back masked too.

    Diagram: Your AI apps send data to Locke, running on your device. Locke detects sensitive data and one-way masks it before it crosses the device boundary to the AI provider. The AI provider's reply crosses back into the device while still masked; Locke 1.0 does not restore raw values; restoration is planned after 1.0. Raw sensitive data never leaves the device boundary.

    Detection (multi-engine) · Masking (one-way) · Return (still masked; restoring the originals is planned after 1.0)

    Try it here. Nothing leaves this page.

    The demo runs in your browser. Open devtools and watch the network tab.

    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 [email protected] 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 detections7 typesNetwork requests: 0

    The AI's replyIn 1.0 it comes back masked.

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

    A simulation of Locke's on-device engine. In Locke 1.0, masking is one-way.

    Click through the full Locke app demo

    The routes Locke screens today

    On supported routes, Locke recognizes AI traffic, including eligible services outside its built-in list. Recognition and screening are separate: detection and masking are supported on the routes below.

    AI websites

    via the browser extension

    ChatGPT
    Claude
    Google Gemini
    Grok
    Perplexity
    more

    Command-line agents

    via the local proxy

    Claude Code
    Codex CLI
    gemini-cli
    more

    IDE assistants, APIs & SDKs

    via the local proxy

    GitHub Copilot
    Official SDKs
    curl & scripts
    OpenRouter, Groq
    Bedrock (four regions)
    more

    Local model servers

    when you point them at Locke

    Ollama
    LM Studio

    Desktop wrapper apps

    recognized and screened by the desktop app

    ChatGPT
    Claude
    Microsoft Copilot

    Coverage depends on the app, platform and settings. Unsupported request types and connections can still bypass screening.

    Locke forChatGPTClaudeGeminiGitHub CopilotCursorall tools

    Locke 1.0 is out. Go get it.

    The whole protection engine, free for one person. Download it for 64-bit Windows or Linux, sign in with a free Sonomos account, and it starts protecting. Teams adds reports and policies at $12.99 a seat. macOS is next.

    Download Locke 1.0

    Or try the interactive demo first.