Skip to main content

    Use AI without handing it your sensitive data.

    Locke is a free desktop app. It spots names, account numbers, patient details and API keys in what you send to ChatGPT, Claude, Gemini and more, and masks them on your device before they leave.

    WindowsLinuxmacOS coming soon

    Free for one person. No card, no time limit.

    Why Locke

    • Stays on your device

      Detection and masking run locally. Your prompts, files and detections are never sent to us.

    • Works past the browser

      AI websites, plus Claude Code, GitHub Copilot, curl and the official SDKs in your terminal.

    • Fails closed

      If Locke can't check a request, it blocks it by default instead of sending it unscreened.

    How it works

    Three steps, and the last one happens on its own.

    1. 1

      Download and sign in

      Install Locke on Windows or Linux and sign in with a free Sonomos account. For AI websites, add the Locke Extension: a separate, free install from your browser's store.

    2. 2

      Use AI the way you already do

      ChatGPT, Claude, Gemini, your terminal agents. No new window to work in and no extra step before you send.

    3. 3

      Locke masks it before it leaves

      Sensitive values become placeholders such as EML_x9k2, so the AI gets the meaning without the data. Greenlight glows green at the edge of your screen so you know it happened.

    Click around the app

    A click-through replica of Locke with invented sample data. Nothing you click leaves this page.

    View as
    Sample data

    Dashboard

    Whether Locke is protecting this device right now, and what it caught. · Monday, October 19

    Protection status

    Running
    Desktop ProtectionUp
    Browser ProtectionConnected
    Detection engineReady

    Configured to monitor 13 AI products.

    The engine is responding. This health check does not confirm that your traffic is being screened.

    Recent activity

    View reports
    email in OpenAI / ChatGPTmasked
    person in Anthropic / Claudemasked
    phone in OpenAI / ChatGPTmasked
    credit_card in Anthropic / Claudemasked
    location in Google Gemini / Vertex / AI Studiomasked

    Detections and first sightings of a surface come from this session's capture feed; policy rows are the proxy's and the extension's own acks of the surface policy; deletions are this device's record. Rule edits are not in this feed — nothing on this device records when a Never flag or a data-type choice changed.

    Risk & compliance posture This device

    Open compliance report
    Evidenced in the available data3 of 7
    Needs your review0 of 7
    Your organization's action4 of 7
    Not assessable from this device0 of 7

    Summarised from the comprehensive requirement checklist over the 19 outbound detection events this session — the same template Reports exports. A tool-generated record, not legal advice. The same summary is shown for every role: Locke has no organization-wide compliance posture.

    23
    Items masked before sending This device
    PII Locke replaced in what you sent since this app opened — across every AI app you used, not one chat. Like the recorded tally, it counts each request's new turn only: the AI tool's own system prompt, the context it injects (saved notes, project files), the tool output it feeds back and re-sent conversation history are masked but not counted. It counts per request rather than per recorded value, so the two figures can differ.
    312
    Sensitive detections recorded (all-time, You typed only) This device
    312 of 947 recorded, per your tally settings — the rest sit in origins you are not counting, or are replayed conversation history.
    Recorded detection occurrences in the selected all-time tally — masking outcomes not included.
    A deliberate subset: everything recorded here was masked, but not everything masked is recorded here.
    13
    AI products enabled This device
    Distinct AI products Locke screens, with at least one hostname enabled in its local surface catalog (27 hostnames in all). Every catalog entry ships enabled, so this is Locke's default list until you change it in Controls — runtime interception is reported separately when available.
    Not counted: Microsoft Copilot in the browser and Web search (7 hostnames), which Locke lists but does not screen — see Controls.
    7 verified · 0 degraded · 20 unverified this session, from captured traffic only.
    4
    Terminal AI tools protected This device
    4 of the 4 terminal tools Locke knows about send prompts to an API host Locke's proxy is still told to screen (Claude Code, OpenAI Codex CLI, Gemini CLI, GitHub Copilot CLI). Locke's surface catalog groups hostnames, not command-line tools, so this mapping is Locke's own — it counts endpoint coverage, not tools installed or running on this machine, and runtime interception is reported separately when available.
    These four figures are not slices of one another: the first is your live capture feed since this app opened, the second is the all-time metrics store narrowed by your tally settings, the third is your local surface list, and the fourth is how many known terminal tools that list still covers. Only the second is stored history. All four are individual figures — this device, under your account. Locke does not show organization-wide totals anywhere yet.
    Detailed status and diagnostics

    Welcome to Locke

    Local diagnostics are ready. A couple of quick steps and you're set.
    2 of 5 done
    Check the detection engineThe detection engine answered a health check. That doesn't confirm your traffic is routed through it.Answering
    2Review configurationSee which AI apps and sites Locke covers.
    3Review local diagnosticsSee the diagnostics available on this device.
    4Try it in the SandboxPaste sample text and review the detector's local diagnostic output.
    See it protect something realDone — Locke recorded its first explicitly protected item from a conforming live producer event.
    Dismissing hides this in this browser.
    Detection engine is answering
    The engine is responding. This health check does not confirm that your traffic is being screened.
    What this check confirms
    Locke's detection engine answered a health check. This check reaches the detection engine directly. It does not pass through the screening path your traffic takes — the system redirect, the proxy, or Locke's guard — so it cannot tell you that any of those are working. If requests are failing or arriving unredacted while this reads green, that gap is where to look. To report this, open Support › Diagnostics and press Copy diagnostics, then send us what it copies.

    Monitoring status This device

    Metrics connected
    Counts are up to date across 46 recorded sessions.
    Counting You typed · occurrences — change what counts
    What the recorded count leaves out
    Locke masks more than it records. Four kinds of masked value are deliberately left out of the total above:
    • Replayed conversation history. A chat client re-sends the whole transcript every turn. A value is recorded once, on the turn it arrives — every later replay is masked again and not counted again.
    • Text the machine wrote. Values written by the AI itself are never recorded. Names, places, organisations and keyword-dictionary hits found in tool output are masked but not recorded.
    • Repeats of a value already counted. The same value in the same position counts once per session. After a large number of distinct occurrences that memory is cleared, so a value seen earlier can be counted once more.
    • Types the metrics store has no column for. A detected type outside its vocabulary is masked but not recorded.
    None of this changes what Locke masks — only what it counts. Your own tally settings then filter that recorded subset further.
    312 sensitive occurrences in the selected all-time tally. Masking success is shown only when the capture pipeline supplies explicit outcomes.

    Live monitoring feed This device

    20 validated detections in the current app-session buffer.
    14:32:08email in OpenAI / ChatGPTmasked
    14:29:51person in Anthropic / Claudemasked
    14:28:34phone in OpenAI / ChatGPTmasked
    14:26:58credit_card in Anthropic / Claudemasked
    last event just now · live tail · processed locally, never leaves your device

    Sensitive data origins (all-time)

    Injected context
    Things your AI tool quietly attaches to every message — saved notes, memory, project files
    not counted39% · 368
    You typed
    Info you actually wrote into your message
    in your tally33% · 312
    System prompt
    The AI tool's own built-in hidden instructions
    not counted13% · 121
    Tool output
    Results from files or commands the AI read, fed back in
    not counted10% · 97
    The AI's replies
    Text the AI generated
    not counted5% · 49
    Much of this content isn't typed by you — the app, its tools, or the AI itself can add it. These are detection/count-only aggregates; they do not establish a masking outcome. Choose what's included under Settings → What counts as a detection. Percentages are the share of all recorded origins, including rows excluded from your tally.

    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

    What Locke catches

    67 types of sensitive data, each with a risk level you can change. A sample:

    Identity
    Names, emails, phone numbers, addresses, dates of birth
    Government IDs
    Social Security numbers, passports, driver licences, NHS and National Insurance numbers
    Financial
    Card numbers, bank accounts, IBANs, routing numbers
    Health
    Medical record numbers, health plan IDs, diagnosis and procedure codes
    Secrets
    API keys, private keys, bearer tokens, database URLs

    Built for Legal, Healthcare, Financial services, Developers and tech and Individuals.

    Greenlight: see Locke work without opening it.

    A long, thin light on the right edge of your screen. The moment Locke masks something, it glows green: your sign Locke caught it, and you can keep going.

    • Green: masked a value or held back an attachment. Red: blocked a request. Amber: something went out unmasked
    • Steady amber when Locke knows protection needs attention, such as a stopped detection engine or capture switched off
    • Clicks pass through it and it never takes focus. An agent's burst of requests is one glow, not a strobe
    • Only a colour reaches it: not what was found, where it went or how many. On Windows it's hidden from screen shares and recordings
    • On by default. Switch it off in Settings › General or Locke's tray menu

    On Windows and Linux with X11. Wayland won't let an app hold the screen edge, so there it's a coloured dot on Locke's tray icon. On GNOME, that needs the AppIndicator extension, which Ubuntu includes.

    On Linux Wayland, the same buttons light the tray icon

    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

    Free for one person

    Teams adds policies and reports for the people you work with.

    Free

    $0

    The whole protection engine, in the browser and the command line. One Privacy summary report a month. No card, no time limit.

    Download Locke

    Teams

    $12.99 per user a month

    Policies pushed to every seat, plus full compliance and technical reports. Admins set the rules and never see content. 30-day free trial, up to 20 seats, US customers only.

    See the plans

    Inside the app

    Every page of Locke, and what it's for.

    What's protected, and what Locke caught.

    The first screen you see when Locke opens.

    • Separate status for desktop protection, browser protection and the detection engine
    • The latest detections and what Locke did about each
    • Four counts, each labelled: items masked before sending, detections recorded, AI products enabled, terminal AI tools protected
    Open the Dashboard in the demo

    You decide what counts as sensitive.

    Pick the AI services Locke screens, the data it masks and what happens to attachments.

    • Protected Surfaces: a switch per AI website or app, screened by the extension or the local proxy
    • Sensitive data: 67 types across identity, government IDs, financial, health, secrets, crypto and legal, each with a risk level
    • Attachments: screen or block by file format. Images and audio can't be checked yet and pass unscreened unless blocked
    Open Controls in the demo

    A report you can hand to an auditor.

    Or to a client, or your own IT team.

    • Compliance reports for Comprehensive privacy, CCPA/CPRA, HIPAA, UK GDPR or EU GDPR
    • Technical reports for IT and security: activity, data types, AI destinations, recommendations
    • Full reports come with Teams. Free includes one Privacy summary a month
    Open Reports as a Teams member

    Every detection, logged without the value.

    Each detection on this device: what Locke did, and where the request was going.

    • Pulse counts for the session, the last hour or the last 15 minutes
    • One row per detection, which opens into its full metadata
    • Filter by direction, surface, data type or outcome. On Teams, export the filtered feed as CSV or JSON
    Open Activity in the demo

    See what the AI would get, before you send.

    Stage a prompt and watch detection mark each sensitive span.

    • Each detection with its type, token and confidence
    • The masked text: each value becomes a one-way token such as EML_x9k2, always the same token for the same value
    • The same engine and disclosure policy as your live traffic, so a test shows what your settings will do
    Run the Sandbox in the demo

    Teams: admins set the rules. They never see content.

    Admins see who has applied the rules. Members start already configured.

    As the admin

    You set policy for every seat.

    • My Team: invite people and see who is active or invited
    • Policies: which AI apps and sites must be screened and which data types are always masked. Nothing changes on a device until you push them
    • Report requests: ask members for a date range and track pending or done. Each member generates the full report on their own device
    Open My Team as an admin

    As a member

    Protected from the first sign-in.

    • Controls becomes a read-only Team policy page showing what the organization set
    • The same 100% local guarantee: admins set the policy, your data stays on your device
    See the member view

    On a team, members cannot pause protection. Admin-approved history clears are logged in the reports each member generates. Prompt content is never collected.

    The details

    A browser extension can't see your terminal. So Locke is a native app.

    Desktop app, not just an extension
    The extension screens AI websites. A proxy on your machine screens command-line and SDK traffic. Ollama and LM Studio join once you point their client at it.
    Your content stays local
    Locke only signs in, checks your licence (account ID and plan, no content) and syncs your settings.
    Attachments
    Text, code, CSV, JSON and log files are screened like your message. PDFs and Office documents are blocked, not sent unchecked. Images and audio aren't screened yet and pass unless you block them.
    Detection
    Checksum-validated patterns plus contextual matching, so a case number is caught even when it doesn't look like one.
    Platforms
    x64 Windows and Linux, available now. macOS is coming soon.

    Questions

    Is the demo on this page the real app?

    A click-through replica of the Locke desktop app, built from its real interface and filled with invented sample data: every name, team and number. Switch between the Free and Teams views, open every page and run the sandbox. Nothing you click is stored or sent anywhere. It shows the interface, not the product's internals.

    What is Greenlight in Locke?

    Greenlight is Locke's activity light: a thin bar on your screen's right edge that glows when Locke acts on a request, so you know it's working without opening the app. Green: it masked something or held back an attachment. Red: it blocked a request. Amber: something went out unmasked. It stays amber while protection needs attention, such as a stopped detection engine. Clicks pass through, it never takes focus, and it only receives a colour, never what was found. It's on by default on Windows and Linux with X11. Turn it off in Settings › General or the tray menu. Wayland won't let an app hold the screen edge, so there the same colours show as a dot on Locke's tray icon.

    What can a Teams admin see?

    Admins manage seats and team policy, ask members for a date-range report, and see which requests are pending or done. No report content reaches the admin: members generate their own. What members type, and what Locke detects for them, stays on their device. Reports cover what was protected, never what people typed.

    Is Locke available now?

    Yes. Locke is available now for x64 Windows and Linux, with no waitlist. macOS is coming soon. Download Locke and sign in with a free Sonomos account.

    What does Locke protect?

    Locke tells AI traffic from other traffic on your device, including eligible services outside its built-in catalog. Recognizing a service does not mean every request is screened. In the browser, the Locke Extension screens AI websites by holding the request inside the page. A proxy on your machine screens command-line and API traffic: Claude Code, gemini-cli, GitHub Copilot, curl and the official SDKs. Local model servers like Ollama join once you point the client at the proxy. Locke also recognizes and screens desktop wrapper apps such as ChatGPT desktop, Claude desktop and Microsoft Copilot desktop. Coverage depends on the app, platform, request type and settings.

    Is anything sent to the cloud?

    Not your content. Detection and masking run on your device. Your prompts, files and detections never go to us or any third party. Locke needs a Sonomos account: it signs in, periodically confirms your licence (account ID and plan tier, no content) and syncs your settings across devices.

    What pricing will Locke have?

    Free for one person, with one Privacy summary report a month. Teams is $12.99 per user a month, one flat price for every seat. It has a 30-day free trial and a 20-seat cap, US customers only. Teams is for teams, companies and organizations needing insight into, and control over, how members use AI. It includes full compliance and technical reports. Enterprise is coming soon: everything in Teams, plus more than 20 seats, custom solutions, white-glove onboarding, a named person at Sonomos and priority support, help with security review and procurement, and custom terms. To hear when it launches, write to [email protected] or use the contact form. Download Locke and sign in with a free Sonomos account.

    Weighing alternatives? See Locke vs Microsoft Purview, vs CASB / SSE, vs self-hosted AI, or all comparisons.

    Locke has its own privacy policy, separate from our website policy. Read the Locke Privacy Policy and the End-User License Agreement.

    Get Locke. It's free.

    The whole protection engine, free for one person: no card, no time limit. Install it on 64-bit Windows or Linux, sign in, and it starts protecting. Teams adds policies and reports at $12.99 a seat. macOS is next.

    Download Locke

    Or try the interactive demo first.