ØCLOAK隠

802.11bf, the WiFi sensing standard, was ratified in September 2025. It reads motion, presence, and breathing through walls using the radio signal a router already sends. Vodafone shipped a consumer version, "Who's Home," in December 2025.

The physics behind it is not new. Every WiFi receiver already computes channel state information to decode its own traffic; sensing just reads that data instead of discarding it. RuView, an open-source implementation, has 76,000 GitHub stars and runs through-wall presence detection on a $9 ESP32.

BLE trackers run on the same asymmetry: cheap hardware, the entire iPhone install base acting as unpaid relays, and a protocol nobody has to authenticate against to use. Facial recognition is a commodity API now too.

Defense against any of this tops out at a handful of $2,000 TSCM sweeps built for corporate clients. Almost nothing exists for a renter who wants to know what's reading their apartment.

The early hardware plan had four devices: a WiFi sensing detector, a broader RF detector, a passive RF-blocking pouch, and a bundle of all three. The roadmap has since narrowed to one lead product: ØCLOAK Guard.

Guard runs on a single ESP32-C3 and does two things. In BLE mode it watches for AirTag, SmartTag, and Tile advertisements, plus the non-compliant clones that skip the anti-stalking rotation schedule entirely. In WiFi mode it watches for the sounding frames (NDP and NDPA) that 802.11bf sensing sessions send, frames that carry no payload encryption because they're control-plane traffic. Combined mode runs both and alerts on either.

Target price is $20 to $25. Manufacturing follows the Pine64 playbook: turnkey builds through Seeed Fusion, 3D-printed enclosures for the first few hundred units, injection molding once around a thousand are committed. The blocking pouches and anti-facial-recognition glasses from the original lineup are deprioritized for now: commodity market for one, a different supply chain for the other.

One legal line runs through all of it: detection and passive shielding are legal everywhere, active jamming is not, in the US or the EU. Guard only listens.

A detector alone tells you about one room. ØCLOAK's network takes sightings from Guard devices and from manual reports, then turns them into a location-tagged map, the way Waze turns individual drivers into traffic data.

Contribution has no account behind it. Each device generates a pseudonymous key that rotates, and location gets rounded to a grid cell before it ever leaves the device, precise enough to be useful and fuzzy enough that a report can't be traced to an address. Reports carry weight: hardware-sourced detections count more than manual ones until other users confirm them.

The plan starts centralized, one API server and one map, because a peer-to-peer layer is real infrastructure work on its own. Decentralization comes later, once there's enough traffic to make community-run coordination nodes worth running.

No VC. Venture money on a privacy company brings growth mandates and an eventual exit, and both point toward monetizing the exact data the product promises to protect. ØCLOAK runs on grants, crowdfunding, and donations instead.

NLnet's NGI Zero fund is the first target: €5,000 to €50,000, low bureaucracy, EU-based, deadline the first of every even month. The Open Technology Fund pays more, $50,000 to $200,000, but it runs on USAGM money, US government funding for an anti-surveillance tool, the same awkward optics Tor has fielded questions about for years. That tradeoff is still open.

Hardware launches go through Crowd Supply over Kickstarter: a 58% success rate against Kickstarter's 30%, and a 100% delivery rate on funded campaigns. Even a project running purely on donations and grants at scale runs a deficit. Signal spends about $36 million a year and outspends its own revenue. ØCLOAK's projections assume the same blend every project in this space ends up needing: grants plus donations plus corporate sponsors, not one clean source.

The transferable asset here is the RF work: channel state analysis, SDR spectrum handling, emitter fingerprinting, sensor fusion, adversarial ML. Consumer privacy hardware is one packaging of that skill. Counter-drone detection is another, and the physics barely changes. A drone is an RF emitter with a control link and a video downlink, plus an acoustic and optical signature. Detect, track, classify.

That direction is under active consideration and it brings real tension with everything above. Defense work ends the open-source and community model through export control and secrecy. Dual-use founders usually get absorbed by the paying customer, and the privacy track goes vestigial unless something structural protects it.

The shape that survives both: one detection engine trained on drone and emitter data recorded first-hand, with two surfaces over it. A networked defense product that pays for the research, and a local-only consumer detector with zero telemetry. Drone detection needs drone data, so the consumer device never has to become a harvesting flywheel. The crowdsourced sensor network sits on hold while that question stays open.

Research and planning. The threat modeling, the hardware bill of materials, and the network architecture are written down. No board has shipped to anyone yet.

The roadmap's first concrete steps: order five ESP32-C3 dev boards, get BLE scanning to detect a real AirTag advertisement, get WiFi monitor mode capturing real sounding frames, and file the NLnet application before its next deadline. That's where the project stands: prototype hardware before a grant, before a crowdfunding campaign, before anything ships.