Just out todayAugmented reality: The AR trainees were faster on the model they trained on, and no better anywhere elseVirtual reality: They matched the exercise dose exactly, and the VR advantage fell below the line that mattersVirtual reality: Both groups used the same headset. One watched a video first and picked up 10.8 more blocks

Cybersecurity/not stated in the record we checked

2,841 medical services with broken authentication, and 94.4 percent with no encryption at all

An internet-wide scan found 2,841 healthcare services with authentication flaws and 94.4 percent of exposed systems with no encryption in transit.

What the paper found

Researchers operated a low-interaction medical protocol honeypot for nine months and ran internet-wide scans across DICOM, HL7 and FHIR on IPv4 and IPv6. They found 2,841 healthcare services with authentication flaws, 94.4 percent of exposed systems with no transport encryption, and 1,373 systems running software with known vulnerabilities including CVSS scores up to 9.8, then ran a coordinated disclosure campaign.

Grade 11 reading level6 min read

What happened

Medical imaging servers leaking patient studies onto the open internet is an established finding. Previous work showed that network segmentation flaws in DICOM systems exposed millions of patient records, and the result has been reported enough times to be familiar.

DICOM carries images. HL7 and FHIR carry the records – admissions, laboratory results, prescriptions, clinical notes – between the systems inside and between hospitals. Nobody had measured their exposure at scale.

A research team ran a low-interaction honeypot for medical protocols for nine months, then scanned the IPv4 and IPv6 address spaces for all three protocols. The work was published at IEEE EuroS&P 2026.

They found 2,841 healthcare services with authentication flaws.

The test

Two complementary methods.

Passive: a low-interaction honeypot for medical protocols, operated for nine months. A honeypot tells you what attackers are looking for, as distinct from what is vulnerable.

Active: large-scale internet scans covering DICOM, HL7 and FHIR across both IPv4 and IPv6 address spaces, identifying healthcare systems and uncovering data leaks due to authentication flaws. The team also scanned for deficiencies in TLS configurations and for known insecure healthcare software.

The combination is what gives the paper its structure: the honeypot establishes attacker interest, the scans establish exposure, and the gap between them is the finding.

The numbers

Measure Result
Honeypot duration 9 months
Scanned for DICOM regularly
Scanned for HL7 or FHIR never
Services with authentication flaws 2,841
Exposed systems without transport encryption 94.4%
Systems with known software vulnerabilities 1,373
Highest CVSS scores observed up to 9.8
Address spaces scanned IPv4 and IPv6
Response coordinated disclosure campaign

The result

2,841 healthcare services were found with authentication flaws – meaning a connection could reach patient data without proving who it was.

94.4 percent of all exposed systems do not support transport encryption. Data moving between those systems travels in the clear.

1,373 systems run software with known vulnerabilities, including some with potential for system takeover and CVSS scores up to 9.8.

The honeypot finding is the one that gives the rest its character. Over nine months the decoy was regularly scanned for DICOM and never once for HL7 or FHIR, which the authors read as indicating that despite their widespread use and importance for patient data security, the security of these services remains underexplored.

The team initiated a coordinated disclosure campaign to improve the security of patient data.

How this compares to what came before

Internet-wide measurement of exposed medical imaging systems has been done several times, and each study has found large numbers of DICOM servers reachable without authentication. That body of work has produced regulatory attention and some remediation.

This paper extends the DICOM work and adds the first large-scale study of HL7 and FHIR. Those protocols are arguably more sensitive: an imaging study is one encounter, while the record stream carries a patient’s history.

The honeypot adds something the scanning literature usually lacks. Measuring exposure tells you what could be attacked. Measuring attacker scanning tells you what is being looked for. Finding a gap between the two is a warning rather than an incident report.

What it means

The 94.4 percent figure is the one to sit with. Transport encryption has been standard practice on the public internet for a decade. Finding that almost all exposed medical systems lack it suggests these are not neglected edge cases but a normal state of deployment.

The honeypot gap is a window and a warning. If attackers are not yet scanning for HL7 and FHIR, the exposure measured here has not been systematically exploited. That window closes the moment someone adds those protocol probes to a scanning toolkit, and publishing this paper arguably accelerates that – which is precisely why the coordinated disclosure campaign matters and why the authors ran one.

For healthcare organisations, the practical question is whether any of your interface engines, record exchange endpoints or imaging systems are reachable from the internet at all. Many will be, through configurations nobody has reviewed in years.

Business ideas from this paper

Healthcare protocol exposure auditing. A scan of an organisation’s own address space for DICOM, HL7 and FHIR endpoints is a short engagement with a concrete deliverable, and almost no hospital has done it.

Transport encryption retrofitting. 94.4 percent without encryption means a large installed base needing TLS termination or tunnelling in front of legacy interfaces that cannot be changed.

Legacy interface gateways. Many HL7 implementations are old and cannot be modified. A gateway that adds authentication and encryption in front of them is a product that fits the constraint rather than fighting it.

Vulnerability monitoring for medical software. 1,373 systems running software with known vulnerabilities up to CVSS 9.8 implies no effective patch monitoring. A healthcare-specific vulnerability feed tied to deployed inventory is a defined service.

What could be wrong with it

Author affiliations were not stated in the record we checked, so we cannot name the institutions behind the work. For a security measurement paper that is a meaningful gap in provenance, and it costs the paper points in our scoring.

Internet-wide scanning counts services, not organisations. One hospital may appear several times behind different addresses, or not at all behind a gateway, so the 2,841 figure does not translate directly into a count of affected institutions.

The scan is a point in time. Exposure changes as systems are deployed and decommissioned.

The paper reports the coordinated disclosure campaign without a long-term remediation rate, so we do not know how many were fixed.

A honeypot measures scanning against that honeypot. Absence of HL7 and FHIR probes there is suggestive and not proof that nobody is looking.

How sure can you be?

We score this 7 out of 10.

It is a peer-reviewed paper at IEEE EuroS&P 2026 – a selective security venue – combining a nine-month honeypot deployment with internet-wide active scanning across two address spaces, followed by coordinated disclosure, which brought third parties into the process.

It loses points because author affiliations were not stated in the record we checked, and because the data are not openly deposited.

The measurement methodology is strong and the disclosure practice is responsible. The provenance gap is unusual for a paper at a venue of this standing and it is what we can and cannot verify that our score reflects.

Do this today

If you run healthcare IT, scan your own external address space for DICOM, HL7 and FHIR endpoints. Finding out whether you are in this dataset takes an afternoon.

If you find an exposed interface, check whether it is doing transport encryption before you check anything else. 94.4 percent of exposed systems were not, and that is the single highest-value fix.

If you are a vendor of healthcare interface software, the vulnerability figure is your customers’ exposure and your reputational risk. 1,373 systems running known-vulnerable software is a patching and communication failure somewhere in the chain.

Source: Measuring Healthcare Data Leaks and Security Flaws at Internet Scale, IEEE European Symposium on Security and Privacy (EuroS&P) 2026, July 2026. arXiv:2607.04965 · doi.org ().

Just Out Tech explains new research in plain language. This article was drafted with AI assistance and checked by a human against the original source.

What to remember
  • First large-scale study of HL7 and FHIR services, the protocols that carry health records between systems.
  • A nine-month honeypot was regularly scanned for DICOM and never for HL7 or FHIR.
  • 2,841 services with authentication flaws; 94.4 percent of exposed systems without transport encryption; 1,373 with known vulnerabilities up to CVSS 9.8.
  • The team ran a coordinated disclosure campaign rather than just publishing the numbers.
Where this really is

The scans covered real healthcare services reachable on the public internet across IPv4 and IPv6, with a nine-month honeypot deployment alongside.

Questions people ask

What are DICOM, HL7 and FHIR?

DICOM is the standard for medical imaging. HL7 and FHIR are the standards for exchanging health records between systems - admissions, lab results, prescriptions, clinical notes.

What does the honeypot result mean?

The researchers ran a decoy medical server for nine months. It was regularly scanned for DICOM and never for HL7 or FHIR, which suggests attackers have not yet turned their attention to the protocols carrying the records.

What is CVSS 9.8?

A vulnerability severity score close to the maximum of 10. Scores in that range typically mean remote exploitation with no authentication required, often leading to full system takeover.

Did the researchers tell anyone?

Yes. The paper reports a coordinated disclosure campaign initiated to improve the security of patient data, which is the responsible practice and not universal in measurement research.

About the author

Mark Alex

Mark Alex is the founder and Managing Director of Real Biz Digital, a technology company operating out of Nairobi since 2018. He works in agentic AI and the Model Context Protocol, AI governance, enterprise software architecture and cybersecurity. He holds an MSc in Mechatronical Engineering from Obuda University in Budapest and a BSc in IT, Forensic Technology and Cybercrime, from USIU-Africa in Nairobi, and has published IEEE conference research on an AI-powered digital twin for greenhouse systems. He is the author of seven books. Between 2020 and 2024 he mentored more than 200 university students and interns in Nairobi. He writes every Just Out Tech article from the original research paper.