Live The Lab is open — real Canary firmware, running in your browser Try it →

Privacy-first, open-source cameras

Three things. One witness.

Home
use cameras to:
detect events
not people|

SecuraCV runs on your own hardware — no cloud, no subscription — and turns cameras into witnesses. It outputs events like "package at front door", never faces, plates, or identities. Not as a setting: the surveillance code doesn't exist.

This is the output. All of it.

Notice there are no images on this site. That's intentional. The system outputs semantic events — text that describes what happened. The surveillance data on the right? That code doesn't exist. Not disabled. Not hidden. Missing.

Event Output
HOME
Not Captured — No Code Exists
facial_identity:███████
race:███████
gender:███████
age_estimate:███████
body_description:███████
license_plate:███████
gait_signature:███████
tracking_id:███████

Your robot vacuum needs to see. Your doorbell needs to see. Your elder care system needs to see. None of them need to identify, track, or remember faces.

The cameras of the future will have eyes — they just don't need to surveil.

Three things. One witness.

When Steve Jobs unveiled the iPhone, he fused three things you already owned — an iPod, a phone, an internet communicator — into one. A Canary does the same, with one twist.

It sees

A camera

The job Ring and Nest sell.

It watches the doorway — but only a sentence leaves the sensor, package at front door, never a face or a plate.

It proves

A notary

The job a signature does.

Every event is Ed25519-signed and hash-chained, so tampering doesn't get blocked — it gets seen, verifiable offline.

It protects

A vault

The job a two-key safe does.

The rare frame that matters is sealed — opening it takes a quorum of trustees. No single person, not even us.

Camera
Notary
Vault
One witness

The twist: Jobs' three helped each other. These three were always enemies — a camera kills privacy, privacy kills evidence, evidence means a stranger's cloud. So we stopped making you pick two. See how it fits together →

Made for people who want cameras, not surveillance

Three kinds of people use SecuraCV today. If you're one of them, there's a path in — and if you're someone else entirely, find your path: every door, sorted by what you came to do.

🏠
Start here — best supported

Home Assistant & homelab users

You have RTSP cameras and a Pi. Frigate keeps recording as usual — SecuraCV runs alongside it, adding the privacy boundary and a tamper-evident witness log. No subscription, ever.

Install in 5 minutes →
⚖️

People who need a record no one can quietly rewrite

Tenants, journalists, activists, abuse survivors. Every event is signed and hash-chained — evidence of what was recorded, where any later edit, by anyone including you, becomes detectable.

See how it's enforced →
🔧

Builders & tinkerers

Cheap ESP32 boards become independent "Canary" witnesses — camera, radar, and WiFi-sensing variants with printable enclosures. Try the real firmware in your browser before you solder anything.

Meet the Canaries in the Lab →

Privacy by absence, not by setting

Most camera software can be configured for privacy. This software lacks the code to violate it. There's no setting to turn off — the capability was never written.

Motion & object events

"Package detected at front door"

Face recognition

No code for this exists

Cross-camera tracking

No code for this exists

Solo video export

Requires multiple trustees to approve

Three intentional constraints

This isn't a feature gap — it's architecture. Every limitation is a deliberate choice.

Events only, not video

The output is a log of what happened. There's no API to stream or export continuous video.

No biometric code

Face, license plate, and gait recognition aren't disabled — the code was never written.

Multi-party video access

Raw footage requires approval from multiple trustees. No one person can extract video alone.

Why build it this way?

This wasn't an accident. Every missing feature is a deliberate design choice.

🔒 Settings can be changed

Every camera system has a "privacy mode." Every one can be turned off by someone with admin access — an employer, a landlord, a government, a hacker.

If the code exists, it can be enabled. We removed the code.

🤖 Machines are getting eyes

Robot vacuums, delivery drones, autonomous vehicles, elder care systems — they all need to see. That doesn't mean they need to identify.

Seeing and surveilling are different capabilities. We built only the first.

⚖️ Today's data is tomorrow's evidence

Video captured for "security" becomes evidence in lawsuits, divorce proceedings, immigration cases, insurance disputes. Footage intended to protect can be weaponized.

The safest data is data that doesn't exist. We output events, not footage.

🏛️ No single point of failure

When one person can export video, you have one person to bribe, coerce, or compromise. Multi-party approval means no single actor — not even the system owner — can extract footage alone.

Trust is distributed, not concentrated.

Same camera. Same question. Different futures.

Watch how "is the coffee ready?" becomes something else entirely.

Traditional System
Full Capture
Break Room Camera
System Output
[09:14]motion_detected zone:coffee_area
[09:14]face_identified: Sarah Chen (94%)
[09:14]badge_correlation: EMP-4821
[09:15]coffee_pot_lifted duration:12s
[09:15]dwell_time: 47s added_to_profile
30-day Storage
847 face events 12.4 GB video
witness-kernel
Events Only
Break Room Camera
System Output
[09:10]coffee_pot: EMPTY
[09:10]motion: zone_A activity
[09:20]coffee_pot: BREWING
[09:30]coffee_pot: READY
[09:30]notification: sent_to_channel
30-day Storage
2,341 events 4.2 MB total

Six months later, a policy changes...

"We need to identify who's been using the break room excessively."

Day 1

HR requests usage report

"Generate a list of employees who spend more than 15 minutes in the break room daily."

Traditional
Report generated: 23 employees flagged, with timestamps, durations, and photos
witness-kernel
Cannot comply. No identity data exists to query.
Day 30

Legal requests footage

"Lawsuit filed. Preserve all break room footage showing employee interactions."

Traditional
6 months of footage preserved. All faces, conversations, behaviors now legal evidence.
witness-kernel
Event log preserved. Shows coffee/motion patterns. No footage to subpoena.
Day 90

New vendor integration

"Our wellness platform wants camera data to track employee stress patterns."

Traditional
API enabled. Facial expressions, posture, interaction frequency now shared with third party.
witness-kernel
Nothing to share. System outputs "coffee ready" — not biometrics.
Day 180

Retroactive analysis request

"Apply our new 'productivity scoring' algorithm to the last 6 months of footage."

Traditional
Analysis complete. Employees scored and ranked by break room behavior. Used in performance reviews.
witness-kernel
Impossible. New rules cannot be applied to historical data. Forward-only by design.
☕ Did you know?

The first webcam just watched a coffee pot.

In 1991, Cambridge researchers pointed a camera at their break room coffee maker. The only question: is it ready yet?

Thirty years later, that same question requires facial recognition, behavioral tracking, and cloud storage. We think the original idea had it right.

Different tools for different needs

This isn't a replacement for existing systems — it's a different category.

Consumer cloud video with E2E encryption

Face recognition Yes
Object detection Yes
Video export Single user
Hash-chained log No
Open source No
Strong encryption, but identifies faces. Any account holder can export clips.

Open-source local NVR for Home Assistant

Face recognition Via add-ons
Object detection Yes
Video export Single user
Hash-chained log No
Open source Yes
Excellent local-first NVR. Anyone with access can export full video.

Different goals, not better or worse

HomeKit and Frigate are excellent at what they do. If you need face recognition, plate reading, or easy video export — use them. Witness-kernel exists for cases where those capabilities are liabilities, not features.

See the full comparison — cost, quality, features & privacy vs. Ring, Nest, Eufy, Reolink, UniFi and Frigate, plus Canary kit pricing →

How privacy is enforced

Six mechanisms that design surveillance out — enforced by hardware and code, not by policy.

I

Frame data is private

Raw pixels are inaccessible. Analysis modules only receive abstract representations.

II

Export requires multiple approvals

No single user can extract video. A configurable quorum of trustees must authorize.

III

Timestamps are coarse

Events are bucketed to 10-minute windows. Precise timing isn't stored.

IV

Log is tamper-evident

Each event includes a hash of the previous. Modifications break the chain.

V

Access tokens are single-use

Emergency access tokens work once, for one time window, then expire.

VI

Rules apply forward only

New detection rules cannot be applied to historical data. No retroactive surveillance.

Data flow

Data Flow
Camera ──▶ RawFrame ──▶ InferenceView ──▶ Module ──▶ SealedEvent
            (private)    (no pixels)               (hash-chained)
               │
               ▼
         FrameBuffer ──▶ BreakGlass ──▶ VaultEnvelope
           (30s max)       (N-of-M)       (if authorized)

Don't trust us — check the code

Every claim on this page can be verified by reading the source.

1

Verify the log can't be tampered with

Run the verification tool to confirm the hash chain is intact.

cargo run --bin log_verify -- --db witness.db
2

Verify frame data is inaccessible

In src/frame.rs, confirm the raw bytes have no public getter.

3

Verify export requires quorum

In src/break_glass.rs, confirm approval counting before token issuance.

Don't read code? Start here. A plain-language walkthrough of what the kernel actually does and how the cryptography really works — sign a real claim, then forge it; break a real hash chain and watch the math catch you. No install, no cloud, nothing to trust but the math.

🧠 See exactly how it works 🔐 Break the glass yourself ✍️ Watch a claim get signed — then forge it

What's built, what's in progress

This is prototype software. Here's where we are.

Frame isolation types

Hash-chained event log

Break-glass quorum

Event contract enforcement

Cryptographic signatures

Encrypted vault envelopes

RTSP video ingestion

WASM module sandboxing

Try it before you build it

No hardware, no install, no sign-up. The Lab runs the real device firmware in your browser.

Live firmware

Meet your Canary before you build one

The display devices run their actual shipping firmware — same faces, same mood engine, same Ed25519 signature checks — compiled to WebAssembly. Touch it, break its network, watch it stay honest. Everything runs locally; nothing phones anywhere.

Open the Lab

Meet the Canaries

Every display device boots its real firmware behind the glass.

Find your Canary

Four questions match you to a device and its printable enclosure.

Enclosure lab

Spin real STLs in your filament color, straight from the OpenSCAD sources.

Build it

Bills of materials with real part numbers and honest totals.

The fleet is still hatching

No fake countdown, no invented backer count. Just what's actually shipped, what's actively being built, and a real place to say "I want one."

2 of 4 designed Canaries shippedWatch Station & Canary Dash are next out of the shop
Counted from the same catalog that runs the store — see the receipts in the open roadmap.
In development

Watch Station & Canary Dash

The bedside glance and the household poster — both running their real, shipping firmware in the Lab today, hardware batches next.

Try the live firmware → Join the waitlist →
Concept · config-only

Canary Vision Pro

Starlight low-light imaging and on-device sound classification for the one spot in your fleet that needs a flagship sensor — no new firmware required.

See how it plugs in →
Concept · in design

The Maker Corps

A community that builds witnesses for people who can't, holds no data, and runs no backend. Three doors are open today; the machinery connecting them is what's still being drawn up.

See the plan →

Read the source. Verify the claims.

Everything on this page is checkable. We're not asking for trust — we're asking for review. Or skip the reading and poke the running firmware yourself.