Public policy proposal · August 2026 · Unclassified
Mandatory cryptographic attestation for every aircraft in American airspace — civil, commercial, law enforcement, and federal. Verify aircraft identity, operator authorization and mission scope separately. Keep every aircraft visible; investigate uncertain identity and unsafe behavior through authorized human review.
The letter to the President and the full technical proposal are published as PDFs at
./docs/. The complete text of both is reproduced on this page, so nothing here depends on a
download resolving.
This extension adds safeguards to the existing Sovereign Clean Sky proposal. The working browser lab now tests signatures, aircraft/operator/mission separation, revocation freshness, signed audit evidence and simulated asset reports. Hardware protections and an operating national registry remain unimplemented.
Controlling clarification: identity, authorization and behavior are separate findings. Authenticated aircraft remain visible. Unverified aircraft require evidence review; missing credentials do not establish nationality, hostile intent or permission to intercept, seize or forfeit property. Earlier language below about automatically subtracting or capturing aircraft is superseded by this clarification. The August letter and downloadable PDFs remain historical documents.
Three identity tiers: government and police use agency-issued credentials and restricted attribution; commercial fleets use organization and aircraft credentials; personal drones use manufacturer-provisioned credentials and secure ownership transfer. Manufacturer provisioning and national mandates are proposals, not existing BB2G authority.
Government privacy: private encrypted mesh and restricted identity resolution protect sensitive records. They do not guarantee invisibility to radar, cameras or radio observers, or override applicable Remote ID obligations. Required deviations need applicable authorization. See FAA Remote ID guidance.
| Gap | Required protection | Limit |
|---|---|---|
| Stolen genuine drone | Authorized operator activation; hardware-bound aircraft proof; stolen-device revocation. | Revocation freshness must remain visible during outages. |
| Copied or replayed credential | Fresh challenge or protocol freshness checks, signed proof of possession, and accepted issuer chain. | A signature alone does not stop replay. |
| Compromised operator account | Strong authentication, device binding, scoped permissions, recovery and revocation. | A legitimate account can still be abused. |
| Authorized operator misuses a flight | Separate aircraft, operator and mission authorization; keep behavior monitoring active. | Verified identity never removes a contact from safety review. |
| Ownership transfer | Revoke previous control rights, verify the transfer, enroll the new owner, preserve custody history. | Aircraft identity and ownership history remain distinct. |
| Government insider abuse | Agency-specific trust roots, mission limits, least privilege, independent review and audit. | Restricted attribution protects sensitive information; it does not eliminate oversight. |
| Compromised mesh node | Mutual authentication, separate agency compartments, scoped keys and node quarantine. | Compromise of one node must not expose the fleet. |
| Offline or stale revocation | Short-lived authorization, cached signed status with age, explicit offline limits. | Missing updates mean status unknown or stale, not safe by default; flight-safety behavior needs separate review. |
| Wrong identity-to-track association | Correlate signed messages with independent observations and show uncertainty. | A valid message does not prove it belongs to the observed aircraft. |
| Issuer or registry compromise | Protected signing keys, dual approval for sensitive changes, independent signed checkpoints, recovery drills. | Trust-root updates and emergency revocation require accountable governance. |
Scroll the table sideways to see all columns.
Preserve the serial number. Research a supplementary manufacturer-bound credential, forensic marking and tamper evidence, with recovery access limited to authorized parties.
A supplementary internal marking or passive tag could help identify a recovered firearm when a visible marking is damaged. It must be evaluated by qualified manufacturers and forensic examiners. Hidden does not mean undiscoverable, unremovable or unclonable. No placement, machining or retrofit procedure is specified here.
Required serial markings remain in place. This research supplements them; it does not replace or conceal them. See ATF firearm identification guidance.
A separate powered module could sign fresh status messages and report supported tamper events. A cryptographic hash describes data integrity; it does not emit a pulse, detect damage or locate an object. Signatures require separate confidentiality and replay protections. See NIST digital signature guidance.
Remote alerts need working power, sensing and a delivery path. Location additionally needs a location source and usable communications; passive tags alone do not provide continuous remote tracking. Battery loss, shielding, damage and outages can prevent immediate reporting.
This research does not control discharge, modify firing mechanisms, promise remote disabling or establish a public gun-location database. It remains separate from aircraft operations and election or child-safety records.
Proposed BB2G Asset Integrity program: authorized issuers manage credentials; an incident desk reviews alerts; independent auditors verify issuance, changes and recovery access.
Change control: sensitive credential, trust-root and ownership changes require authenticated requests and two-person approval. Preserve before-and-after evidence, signed versions, reason and approvers. Compare records with independently retained checkpoints so an administrator cannot quietly rewrite the entire history.
Incident flow: detect or receive a report → validate evidence and freshness → record a review flag → notify authorized parties when delivery is available → human decision and documented resolution. Log dispatch, delivery and acknowledgment separately.
Privacy and accountability: separate drone and firearm records; restrict identity and location access; record purpose and authority; set retention periods; permit correction and appeal; audit issuer compromise and false theft reports. No automatic nationality classification or enforcement action.
Acceptance work: test stolen aircraft, replay, account compromise, stale revocation, mission misuse and a compromised mesh node. For firearm research, independently test mark recovery, tag removal or substitution, sensor coverage, legitimate service, low power, no signal, false theft reports and location freshness. Publish measured detection and delivery times, missed events and false alarms. No “instant,” “tamper-proof” or “untraceable” claim without evidence.
Open the working verification lab · Read the complete design requirements and acceptance matrix.
Require every drone in American airspace to carry a cryptographic identity it must continuously prove. Then anything that cannot prove itself is, by definition, the thing you are looking for — and you go get it, intact, and put it in an evidence locker.
This is not a detection program. Detection is the part everyone already spends money on, and it is the part that does not scale — you cannot inspect your way through a sky full of anonymous aircraft. This proposal inverts the problem. Instead of asking sensors to find the bad drone in the noise, it removes the noise: every legitimate aircraft becomes provable, so every illegitimate one becomes conspicuous.
Every component below already exists. The cryptography is a published IETF standard. The legal authority is enacted federal law with a 2031 sunset. The capture hardware has thousands of operational recoveries. What does not exist is the wiring between them — and one new layer, described in section 06, that nobody is building.
Since September 2023, most drones in American airspace must broadcast identification and location in flight under 14 CFR Part 89. Every registered aircraft, every Part 107 operation, every recreational drone over 0.55 pounds (FAA Remote ID).
Those broadcasts are plain text with no authentication. Researchers describe Remote ID as “critically incomplete from a security perspective,” open to spoofing, relay, and replay (arXiv). It carries the same structural weakness as ADS-B — clear-text messages, no mandatory encryption or authentication (peer-reviewed survey, Linköping University). Working spoofers are published on GitHub and run on a Raspberry Pi (Cyber-Defence Campus).
So today, “it is broadcasting an ID” proves nothing. A hostile operator invents an identity and becomes indistinguishable from a delivery drone — or floods the picture with a thousand fictional aircraft and destroys the operator's ability to trust any of it.
The IETF's DRIP working group exists to make Remote ID trustworthy (charter) and has already shipped it.
RFC 9575 states the problem outright: existing Remote ID regulations and standards “do not address trust.” The token, the person-and-organization hierarchy, the endorsement chain, the anti-replay proof — all written, standardized, and mandated nowhere.
The gap is not invention. It is adoption.
Counter-drone work today is a needle-in-a-haystack problem. Attestation converts it into a set-difference problem: what the sensors see, minus what can prove itself. Remaining identity gaps require review. Verified identities remain subject to mission and behavior checks. This illustrative demo includes a verified aircraft requiring mission review. Press the control below.
The deterrent effect does not come from firepower. It comes from the end of anonymity. An operator who knows that flying unattested is itself the trigger — and that the airframe will be recovered rather than destroyed — is facing the loss of the hardware, the flight log, the payload, and the launch point. That is a different calculation than risking a shootdown that conveniently erases the evidence.
The statute already authorizes exactly this sequence. Stopping deliberately short of force is what makes the program legally clean, operationally superior, and politically survivable.
| Subsection | Authorized action | This proposal |
|---|---|---|
| (b)(1)(A) | Detect, identify, monitor, and track, without prior consent | Map |
| (b)(1)(B) | Warn the operator, including through the Remote ID broadcast channel | Warn |
| (b)(1)(C) | Disrupt control of the aircraft | Skip |
| (b)(1)(D) | Seize or exercise control of the aircraft | Capture |
| (b)(1)(E) | Seize or otherwise confiscate the aircraft | Forfeit |
| (b)(1)(F) | Use reasonable force to disable, damage, or destroy | Never |
Scroll the table sideways to see all columns.
Counter-drone authority lives in 6 U.S.C. §124n, which shields agencies from statutes that would otherwise make interception illegal — the Wiretap Act, the Computer Fraud and Abuse Act, the Aircraft Sabotage and Aircraft Piracy Acts. The SAFER SKIES Act, enacted December 2025 in the FY2026 NDAA, extended that authority to September 30, 2031 and for the first time gave state, local, Tribal, and territorial law enforcement mitigation power (Congressional Research Service). A DOJ/DHS Interim Final Rule became legally effective July 1, 2026 (Crowell, Homeland Security Today), with Tier 1 detection certified online and Tier 2 mitigation requiring in-person training at the FBI's National Counter-UAS Training Center.
The doctrine is not a novel legal theory. Subsections (D) and (E) authorize “seize or exercise control” and “seize or otherwise confiscate” — capture and forfeiture, written into federal law almost line for line.
A drone flying a GPS-waypoint mission carries its profile onboard and transmits nothing. No datalink to sever, no uplink to overpower, no signal to intercept. Both sides in Ukraine adapted commercial drones to fly RF-silent for exactly this reason.
Net capture is spectrum-independent: a propeller catches the mesh, motors stall under load, lift is lost. It works against autonomous, fiber-tethered, and swarming systems (technical guide).
A netted airframe arrives intact with SD cards, flight logs, and payload available. Shooting a drone down ends one incident; capturing it starts an investigation — operator, launch point, route, target, payload, and if it is part of a coordinated effort, the rest of the network.
A drone that disappears in a fireball is deniable. A drone sitting in an evidence locker with its flight log intact is not.
4,500+ captures
The Fortem DroneHunter F700 does not jam and does not spoof. It pursues, fires an expanding tethered net, and tows the capture to a forensics recovery point — or lowers it under a parachute. Across three continents (UAS Feed).
Onboard AESA radar
Guidance is self-contained, with an 85% first-shot capture rate and a second shot usually ready to follow. It has recovered platforms larger than itself, including the Orlan-10 and the Shahed-136.
Not a single vendor
ParaZero DefendAir is shipping for critical infrastructure and ARGUS A1-Falke is adding lidar for terminal precision. Net interception has a supplier base, not a sole source.
This is simultaneously aviation infrastructure and national security infrastructure — which is exactly why it has never been built. Layers 1 through 3 belong to the FAA. Layers 4 through 6 belong to DHS, DOJ, and DoD. The handoff is where the program lives or dies.
Issue cryptographic identities to aircraft, operators, and organizations. Maintain endorsement chains and revocation.
The aircraft continuously proves possession of its private key over the air, signing unpredictable data.
Collect, correlate, and distribute the airspace picture to authorized parties across service suppliers.
Fuse radar, RF, EO/IR, and acoustic tracks. Match what sensors see against what the sky claims. Flag every mismatch.
Intercept, net, tow, and seize the airframe intact. Preserve chain of custody for prosecution.
Encrypted, air-gapped interceptors that collect forensic evidence in flight, coordinate over a closed mesh, and drive pattern-of-targeting analysis.
Each one is an encrypted, air-gapped flying forensic node that carries its own attested identity, talks only to its own fleet over a closed mesh, and begins documenting the target the moment it launches. By the time the net fires, we already have the profile. The capture just confirms it.
| Collection | What it yields |
|---|---|
| RF fingerprint (raw I/Q) | Hardware-level identity of the transmitter, independent of any claimed ID |
| Control-link characterization | Protocol, hopping pattern, whether the aircraft is RF-silent and autonomous |
| EO/IR imagery | Airframe type, payload, modifications, markings |
| Acoustic signature | Motor and propeller class — useful when RF is absent entirely |
| Kinematic track | Launch bearing, route, loiter points, target dwell time |
| Attestation transcript | Exactly how and when identity verification failed — the legal predicate |
Scroll the table sideways to see all columns.
The RF fingerprint is the load-bearing one, and it is real science. Fingerprinting extracts nuanced but sufficiently detailed characteristics from a received signal (arXiv). Ensemble models like DrIfTeR detect and classify drones purely from RF fingerprints; wavelet analytics have been benchmarked for the same task; CrossRF attacks the generalization problem directly. There is even prior art for the inverse — AirID injects a deliberate RF signature into transmitted I/Q samples to mark authorized aircraft.
Why this is a genuine advance: a cryptographic token can be absent, but an RF fingerprint cannot be. A drone that refuses to identify itself still radiates a hardware signature. Pair attestation — what it claims — with fingerprinting — what its transmitter is — and you can recognize the same physical airframe across separate incidents weeks apart, even if it never broadcasts a valid token once. That is what turns individual captures into a case file.
Fully air-gapped end to end. Onboard encrypted vault, mesh-only coordination, recovery to an on-premise forensic lab. No cloud dependency of any kind. The configuration for contested environments.
After recovery, evidence crosses a data diode into an Azure confidential-computing enclave. The hardware is physically incapable of bidirectional transmission — no return pathway, no reverse communication (Nexor, NATO).
Real-time streaming during flight. Faster analysis and remote human review, at the cost of a jammable, detectable, attackable dependency. Acceptable for a stadium perimeter. Never the default — falls back to Tier A the instant the link degrades.
The destination side exists at government classification levels today. Azure Confidential Computing reached general availability on AMD SEV-SNP virtual machines across all U.S. government data classification levels (Microsoft Azure Government), hardening the environment against the host, the hypervisor, the host administrator, and even your own VM administrator (Microsoft Learn). Microsoft Sovereign Cloud adds data-residency and operational-sovereignty controls. The cloud forensics layer belongs on the far side of a diode — not on the airframe.
This is the most operationally valuable capability in the program, because it addresses a harm that is already happening and already prosecutable.
Drone harassment and stalking are crimes under existing law, and local stalking, trespass, and harassment statutes have already secured charges without waiting on the FAA (DRONERESPONDERS). In Rockingham County, Virginia, a man was charged with two counts of spying using an electronic device, four counts of entering a property to harass using a drone, and two counts of stalking with fear of death or assault (Police1). Practitioners note there is no single drone law for this — ordinary trespass, voyeurism, harassment, and stalking statutes get applied to drone facts, and the civil/criminal line generally turns on intent and repetition (Watkins).
Read that last clause again. Intent and repetition. Repetition is exactly what the Sovereign Fleet produces. A single overflight is ambiguous. Nine overflights of the same residence at the same hour across three weeks, tied to one RF fingerprint, is a stalking case with the element of repetition already proven. Today victims cannot establish that pattern because nobody is recording it.
Link separate incidents to the same physical airframe via RF signature, even with no valid token ever broadcast.
Flag recurring overflights of the same location, clustered by time of day and day of week.
Distinguish transit from loiter, and loiter from directed observation of a specific structure.
Back-track kinematic tracks across incidents to converge on the operator's staging area.
Connect incidents sharing fingerprints, hardware, or launch geometry into one case — whether the target is a person or a substation.
A sealed, hash-chained package per incident and per campaign: attestation failure proof, fused track, fingerprint match, intercept log, custody record.
The analysis profiles drone behavior, not residents. It must never become a standing database of who lives near what. A targeting analysis for a private individual is opened on that individual's complaint or consent, or under a protective order — never as ambient monitoring of a neighborhood. Aggregate pattern analysis runs against protected facilities and critical infrastructure, which have institutional standing to consent.
Build it the other way and you have created the surveillance system this project family exists to argue against — and it will be attacked on exactly those grounds. Victim-initiated, consent-scoped, and audit-logged is both the ethical design and the survivable one.
The expensive phases come last. The decisive one costs a rulemaking and can begin immediately, and the whole schedule is aligned to close before counter-UAS authority sunsets on September 30, 2031.
Direct the FAA to name DRIP (RFC 9374 / RFC 9575) as the authentication layer for Part 89, with a fixed compliance date. No new science, no new appropriation.
Start hereStand up the issuance authority, endorsement chains, revocation, and the Session ID–style privacy model. Include the separate government hierarchy from day one.
One critical-infrastructure site runs the full set-difference picture: attested traffic normal, everything else escalated, adversary spoofers run against the verifier.
Wire attestation failure to intercept tasking. Recover airframes to a forensic lab under hash-chained custody. Prosecute the first cases.
Air-gapped forensic interceptors, mesh coordination, pattern-of-targeting service. Retrofit path closed out before the authority sunset.
Scroll sideways to see all five phases.
Presented without softening. None of the five requires new science, and the first requires no new money.
Name RFC 9374 and RFC 9575 in rulemaking under 14 CFR Part 89, with a fixed compliance date and a published retrofit path. The standards are already Standards Track. This is an adoption decision.
Issuance, endorsement, and revocation for aircraft, operators, and organizations. Public observers see a rotating pseudonym on the existing Remote ID Session ID model; identity resolves only through the registry to authorized parties.
Valid token, restricted attribution, resolvable only by cleared authorities. A blanket carve-out would break the set-difference rule and hand every adversary a category to hide in.
Exercise (D) and (E) before (F). Fund net-capture recovery and forensic intake so every intercepted airframe becomes evidence instead of debris, and set a national chain-of-custody standard for recovered aircraft.
Consent-scoped, audit-logged, profiling drone behavior rather than residents — so that a stalking victim can prove repetition, and a substation operator can prove a campaign.
We will build and publish the working proof at our own expense: a DET issuance authority, a signing aircraft simulator, an independent verifier, and an adversary harness that runs the public Remote ID spoofer against both the current unsigned rule and the DRIP-authenticated one, side by side. Nobody has to take the argument on faith. We will show the unsigned rule collapsing and the signed one holding.
BORN BETWEEN GENERALS, LLC
Operation Angels · BB2G Solutions
The President of the United States
The White House
1600 Pennsylvania Avenue NW
Washington, DC 20500
Re: Sovereign Clean Sky — closing the authentication gap in American airspace
Mr. President,
Every drone flying legally over the United States is already required to announce who it is. Almost none of them can prove it. The Remote ID broadcast mandated under 14 CFR Part 89 is unauthenticated plain text — a license plate anyone can print. Working spoofers are published openly and run on hardware that costs less than dinner. This means that on the most basic question we can ask about an aircraft over a military installation, a substation, a stadium, or a family's back yard — who are you — our answer today is an unverified assertion from the aircraft itself.
I am not writing to report a gap you have not already acted on. Your Executive Order on Unleashing American Drone Dominance pushed this industry forward, and the SAFER SKIES Act extended counter-drone authority through September 30, 2031 and put it, for the first time, in the hands of state and local law enforcement. The authority exists. The industry exists. What is missing is the one layer that makes both of them mean something: proof.
The remedy is already written. The Internet Engineering Task Force has published the standard that makes Remote ID trustworthy — an identifier that is itself a public-key commitment, and an authentication message in which the aircraft proves it holds the private key by signing unpredictable data. It is Standards Track. It is finished. It is mandated nowhere. The gap is not invention. It is adoption.
And adoption changes the shape of the problem, not just its difficulty. Today we ask sensors to find one hostile drone inside a sky full of anonymous ones — a needle in a haystack, and the haystack grows every year. Require attestation and that becomes simple subtraction: what the sensors see, minus what can prove itself. Whatever is left is the entire target list. A clean, attributable sky makes an unattributable aircraft conspicuous. That is the whole idea, and it is why this is infrastructure rather than equipment.
On what happens next, I want to be equally plain, because I believe the current instinct is wrong. We should not shoot these aircraft down. Federal law already authorizes us to seize or exercise control and to seize or otherwise confiscate — capture and forfeiture — before it ever reaches the use of force. Net-capture interceptors have completed thousands of recoveries across three continents. A drone brought down under a net arrives intact, with its flight log, its payload, and its launch point still readable. A drone destroyed in the air arrives as debris and a deniable incident.
Deterrence does not come from the fireball. It comes from the end of anonymity, plus a recovered airframe sitting in an evidence locker. Do that a hundred times and we are no longer defending airspace one incident at a time — we are mapping an adversary's entire program, and we are doing it with evidence that survives a courtroom.
There is one more thing I would ask you to look at, because it is the part nobody is building. Our own interceptors should be encrypted, air-gapped, and attested under the government's own hierarchy — no uplink to jam, no cord to cut, no stolen airframe worth stealing — and they should fly as forensic instruments, not just as nets. They can capture a transmitter's hardware fingerprint, which cannot be forged the way a token can be forged and cannot be withheld the way a token can be withheld. That single capability lets us recognize the same physical aircraft across separate incidents weeks apart, which is how a series of overflights becomes a provable pattern instead of a string of unrelated complaints.
That matters far from any military fence line. In Virginia, a woman was followed over her own property by a drone for months before charges were finally brought. The law she needed already existed; what she did not have was a record. The line between a nuisance and a crime usually turns on intent and repetition, and repetition is a data problem we are simply not solving. Built correctly — victim-initiated, consent-scoped, audit-logged, profiling drone behavior and never residents — this is the difference between telling that woman we are sorry and handing her a case.
Accordingly, I respectfully ask for five things:
The first of those costs a rulemaking, not an appropriation, and it is the one that unlocks the other four. And you do not have to take any of this on faith. My company will build and publish the working proof at our own expense — an identity issuing authority, a signing aircraft, an independent verifier, and an adversary harness that runs the publicly available spoofer against the current unsigned rule and the authenticated one side by side. We will show one collapsing and the other holding. If it fails, we will publish that too.
I am a small business owner in Brandon, Florida, and I build proof-oriented systems because I believe institutions earn trust by making their claims verifiable rather than by asking to be believed. I am asking the United States to apply that same standard to its own sky. The authority is in hand and it expires in 2031. The standard is finished and waiting. What remains is a decision.
Thank you for your time and for your service to our country. I am at your disposal, and everything on the accompanying site — including every limitation of this proposal — is published in the open.
Respectfully,
Krissy Hall
Founder · Born Between Generals, LLC
Brandon, Florida · cleansky.bornbetween2generals.com
This letter is a private citizen's policy proposal. It is reproduced here in full and also published as a signed PDF.
Anything that will be used to attack this proposal is stated here first, by us.
Retrofit is the hard cost. Millions of aircraft already fly with unsigned Remote ID. Signing needs key storage and compute, and Bluetooth message budgets are tight — which is why RFC 9575 required a specialized wire format. A hard cutover would orphan a large legacy fleet.
The government carve-out is the hard politics. The Department of Defense will not publish keys that let anyone geolocate sensitive operations. A separate hierarchy with restricted attribution is the workable answer, but it is a policy fight, not an engineering one.
RF fingerprinting does not generalize cleanly. Accuracy degrades across channel conditions and hardware batches. Fingerprint matches must be treated as investigative leads with confidence scores, never as courtroom-grade identity on their own.
Size, weight, and power constrain the airframe. A small interceptor cannot carry a full forensic lab. Lightweight onboard capture, heavy analysis in the ground enclave. We do not claim onboard deep learning at scale.
Key management is the real program risk. Secure elements, provisioning, rotation, revocation across an intermittently connected mesh, and recovery when an interceptor is lost. This is where sovereign systems usually fail, and saying so is not a formality.
A national registry is a sensitive database. It concentrates identity information, and it must be governed accordingly. The existing Remote ID Session ID precedent is the mitigation we propose, not a reason to dismiss the concern.
Part 108 BVLOS status is unsettled in public reporting. Sources conflict on whether the rule is final. We treat it as pending and do not build any part of this argument on it.
The primary references, collected. Standards and statutes link to the authoritative text, not to commentary.
The working prototype uses classical P-256 signatures. It has not been independently certified for post-quantum protection. This status describes the design; it is not a finding that the site or a financial account has been compromised.
Sovereign Clean Sky is owned by Born Between 2 Generals LLC (BB2G). The verification lab is free to use. No quantum-monitoring subscription is required.