A network that shares threats,
never patients.
One hospital's attacker becomes every hospital's warning. The only thing that crosses the boundary is an IP address, so a patient record never can.
De-identified threat federation turns each participant's local attack data into shared defense across the RankShield Network. The only indicator that federates is an IP or CIDR range, validated by net.isIP. Protected health information has no path to the mesh by construction, and every report is sealed to the ledger for attribution.
Every hospital
defends alone.
When an attacker probes one hospital, that hospital learns something the rest of healthcare urgently needs — the address the attack came from. But today that lesson stays trapped inside one network. The same actor moves to the next site and starts fresh, unknown. Each institution rediscovers the same threats in isolation because the one signal worth sharing — a suspect IP — has nowhere trustworthy to go. Federation gives it somewhere: a shared network where one site's discovery becomes everyone's defense.
The only thing
that travels is an address.
The federation path accepts exactly one kind of value: an IP address or CIDR range, validated by net.isIP before anything publishes. There is no field for a name, an MRN, a diagnosis, or a payload. Because the only value that can travel is a network address, protected health information cannot reach the mesh — not as a matter of policy, but of construction. PHI-impossible, not merely PHI-free.
Three gates
to federate.
A report only publishes to the RankShield Network if the reporter clears three gates: a verified RFC 9421 identity, a BAA on file, and an explicit federation opt-in. Miss any one and the report is still sealed to the ledger for attribution — but it does not federate. The barrier to entry is what keeps the shared signal accountable: every published indicator traces to a real, opted-in healthcare entity.
No source
trusted alone.
The barrier is not the last check. The category medical is deliberately not in the mesh's trusted set, so high and critical indicators are downgraded unless independently corroborated by other sources. That corroboration gate is a third layer on top of identity, BAA, and opt-in. One compromised or mistaken reporter cannot poison the shared signal, because no single source, including this one, is trusted on its own word.
Share threats.
Never patients.
Clear the barrier once — verified identity, BAA, opt-in — and every attacker you see helps defend every other participant, while theirs help defend you. IP and CIDR only. Sealed for attribution. PHI-impossible by construction.
What is de-identified threat federation in healthcare?
De-identified threat federation in healthcare is the sharing of a single narrow, de-identified signal — a suspect IP address or CIDR range — across a network of hospitals and clinical vendors, so that one participant's discovery of an attacker becomes every participant's warning, without any patient data ever crossing the boundary. The premise is simple and the discipline is strict. When an attacker probes one hospital, that institution learns the network address behind the attempt; that address is exactly the kind of signal the rest of healthcare needs and exactly the kind of signal that, today, stays locked inside one network. RankShieldMD federates that one value and nothing else. The indicator that publishes to the RankShield Network is an IP or CIDR range, validated by net.isIP before it can travel; there is no field on the mesh path for a name, a medical record number, a diagnosis, or a payload. Two principles govern the design, and we hold to both honestly: only an address federates, so protected health information is impossible by construction rather than filtered by policy, and every report — whether it publishes or is blocked — is sealed to the ledger for attribution. It supports your compliance posture; it does not by itself make you compliant.
How is it PHI-impossible, not just PHI-free?
Because the difference is architectural, not procedural, and in healthcare that distinction is the whole point. “PHI-free” describes a system that lets many kinds of data in and then filters protected health information out — a policy that can be misconfigured, bypassed, or simply fail under load. “PHI-impossible” describes a system where protected health information has no path to reach the shared mesh in the first place. In the RankShield Network, the only value that federates is an IP address or CIDR range, and it must validate as an IP through net.isIP before it can publish. There is no PHI-shaped field to populate. A name is not an IP; a medical record number is not an IP; a clinical note is not an IP — so none of them can be published as an indicator, no matter what a caller intends or misconfigures. The de-identification is enforced by the shape of the data the path accepts, not by a downstream scrubber hoping to catch everything. That is why we are careful with the word: this is PHI-impossible by construction, and we describe it precisely rather than overclaiming. It shrinks the surface where a mistake could expose a patient, because the surface for that mistake does not exist on the federation path. And because even blocked reports are sealed to the ledger as digests and de-identified indicators, not patient data, the record of who reported what is accountable without ever holding anything protected.
What is the participation barrier and why does it exist?
The participation barrier is a deliberate barrier to entry: a report only federates to the RankShield Network if the reporter clears three gates, and it exists because a threat network is only as trustworthy as the parties inside it. The first gate is a verified RFC 9421 identity — the reporter must prove who it is by signing its request, not merely assert it. The second is a Business Associate Agreement on file — the reporter must be a real, contractually accountable healthcare entity. The third is an explicit federation opt-in — participation is a choice the organization makes, never a default it stumbles into. Clear all three and the indicator publishes to the mesh. Miss any one and the report is still sealed to the ledger for attribution, so there is an accountable record of who reported what and when, but it does not federate. The reason for raising the cost of joining is precisely to keep the shared signal clean: every published indicator traces back to a verified, opted-in institution that has agreed to participate under a BAA. A threat network with no barrier fills with noise and bad actors; a threat network with the right barrier fills with accountable participants whose signals are worth consuming. The barrier is not friction for its own sake — it is the thing that makes federation trustworthy enough to act on. We claim only that: it gates who can publish, and it seals everything for attribution. It does not, on its own, make any organization compliant.
How does mesh corroboration prevent poisoning?
By refusing to trust any single source — including healthcare itself — on its own word. The participation barrier controls who can publish, but a determined or compromised participant could still try to elevate a false indicator. Mesh corroboration is the answer, and it works as a third layer on top of identity, BAA, and opt-in. The category medical is deliberately not in the mesh's trusted set. That means a medical-origin indicator does not inherit high or critical severity simply because a medical participant reported it; high and critical indicators are downgraded unless they are independently corroborated by other sources on the network. A single reporter, no matter how well credentialed, cannot unilaterally raise an indicator to a level that forces action everywhere. The severity has to be earned through agreement across independent origins. This is a deliberate humility in the design: the network assumes that any one source can be wrong or compromised, and it structures trust so that no single point of failure can poison the shared signal. It is the same discipline that runs through everything RankShieldMD builds — claim only what independent evidence supports — applied to the moment where a poisoned feed would do the most damage. The result is a network where the barrier keeps unaccountable parties out, and corroboration keeps even accountable parties from being trusted blindly. Together they make the shared indicator something a defender can act on without inheriting one reporter's mistake.
Why the network effect is the moat
Because the value of the RankShield Network is not a feature that can be copied but a network that would have to be rebuilt. Every participant publishes to and consumes from the same mesh: the attackers one hospital discovers become indicators every other hospital can block, and every other hospital's discoveries become indicators it can block in turn. That reciprocity compounds. Each new accountable member — each institution that clears the barrier and contributes corroborated indicators — makes the shared signal richer for every other member, which in turn makes joining more valuable, which draws more accountable members. A competitor can copy an interface in an afternoon; it cannot copy the accumulated, corroborated, barrier-gated signal of a network of participants who already trust and feed it. The moat is the reciprocity itself, disciplined by the two mechanisms that keep it trustworthy: the participation barrier, which ensures every publisher is a verified, opted-in, BAA-covered entity, and mesh corroboration, which ensures no single source is trusted alone. And the entire compounding advantage accrues without any patient data ever crossing the boundary, because the only thing that federates is an IP or CIDR range. We describe the effect precisely: it supports collective defense and grows more valuable with each member, and it does not replace an organization's own controls or make it compliant. The network effect is the moat because it is defense that gets stronger every time someone else joins — and stronger in a way no one can shortcut.
What we are careful never to claim.
PHI is impossible, not just filtered
Only an IP or CIDR range federates, validated by net.isIP. There is no PHI-shaped field on the mesh path, so protected health information cannot reach it by construction. It supports compliance; it does not make you compliant.
The barrier is required, not optional
Federation requires a verified RFC 9421 identity, a BAA on file, and an explicit opt-in. Blocked reports are still sealed to the ledger for attribution. Only barrier-pass reports publish to the network.
No source is trusted alone
The category medical is not in the mesh's trusted set. High and critical indicators are downgraded unless independently corroborated, so one reporter can never poison the shared signal.
Ask RankShieldMD about de-identified threat federation.
What is de-identified healthcare threat federation?
It is the sharing of one narrow, de-identified signal — a suspect IP address or CIDR range — into the RankShield Network, so that when one participant sees an attacker, every other participant can be warned. The only indicator that federates is an IP or CIDR block, validated by net.isIP; nothing else travels. It turns each hospital’s local attack data into shared, corroborated defense without ever moving a patient record.
What actually gets shared on the mesh?
An IP address or CIDR range, and nothing more. The indicator is validated as an IP by net.isIP before it can federate. There is no field on the mesh path for a name, an MRN, a diagnosis, a payload, or any free text. Because the only value that can travel is a network address, protected health information has no path to the RankShield Network by construction.
How is this different from a normal threat feed?
A normal feed is a one-way subscription. The RankShield Network is a two-way membership: every participant both publishes to and consumes from the same mesh. Your defense improves every other member’s defense, and theirs improves yours. That reciprocity, gated behind a real participation barrier, is what makes the network compound rather than merely distribute.
How is it PHI-impossible, not just PHI-free?
“PHI-free” describes a policy that filters PHI out. “PHI-impossible” describes an architecture where PHI can never reach the mesh path in the first place, because the only value that federates is an IP or CIDR that must pass net.isIP. There is no PHI-shaped field to fill. It supports your compliance posture; it does not by itself make you compliant.
Could a mistake leak a patient identifier onto the mesh?
The federation path only accepts a value that validates as an IP address. A name, an MRN, or a note is not an IP, so it cannot be published as an indicator. The de-identification is structural rather than procedural, which is why we describe it as PHI-impossible by construction rather than PHI-free by policy.
Does the ledger contain any protected data?
No. Every report is sealed to the ledger for attribution, but what is sealed is a digest and the de-identified indicator, not patient data. Even a report that is blocked at the barrier is sealed for accountability; none of it carries PHI, so the ledger is useless to anyone who steals it.
What is the participation barrier?
A report only publishes to the RankShield Network if the reporter clears three gates: a verified RFC 9421 identity, a Business Associate Agreement on file, and an explicit federation opt-in. Miss any one and the report is still sealed to the ledger for attribution, but it does not federate. The barrier is what keeps the shared signal trustworthy.
Why require a BAA and an opt-in to federate?
Because a threat network is only as trustworthy as the parties inside it. Requiring a verified identity, a BAA, and an opt-in means every published indicator traces to a real, accountable healthcare entity that has agreed to participate. It raises the cost of joining specifically to keep bad actors and unaccountable noise out of the shared signal.
What happens to a report that fails the barrier?
It is sealed to the ledger for attribution, exactly like a passing report, so there is an accountable record of who reported what and when. It simply does not publish to the mesh. Only barrier-pass reports become shared indicators; everything else stays local and attributable.
How does mesh corroboration prevent poisoning?
The category “medical” is deliberately not in the mesh’s trusted set, so a single medical report cannot elevate an indicator on its own. High and critical indicators are downgraded unless independently corroborated by other sources. That corroboration gate is a third layer on top of the identity-plus-BAA-plus-opt-in barrier, and it is what stops one compromised or mistaken reporter from poisoning the shared signal.
Why is “medical” not a trusted source by default?
Trusting any single origin unconditionally is exactly how a threat network gets poisoned. By keeping “medical” out of the trusted set, the mesh forces medical-origin indicators to earn their severity through independent corroboration rather than inheriting it. It is a deliberate humility in the design: no source, including this one, is trusted on its own word.
Why is the network effect the moat?
Every participant publishes to and consumes from the same mesh, so each new member makes the shared signal richer for all the others. That compounding value is difficult to replicate, because it is not a feature to copy but a network to rebuild. The reciprocity, gated by the barrier and disciplined by corroboration, is the durable advantage.
Is the network effect just marketing?
No. It is mechanical: a report that clears the barrier and earns corroboration becomes an indicator every other participant consumes, and every other participant’s cleared reports become indicators you consume. The value of joining rises with each accountable member, and it does so without any patient data ever crossing the boundary. It supports collective defense; it does not replace your own controls.
Turn one hospital's attacker into everyone's warning.
Clear the barrier once, and every threat you see helps defend the network — while the network helps defend you. IP and CIDR only, PHI-impossible by construction.