Photon Health; Healthcare Highways announce data security incidents, and one of the two breaches points to a risk many health technology companies overlook: the analytics server. Photon Health, a Brooklyn, New York company that runs a prescription network, learned on August 21, 2026, that attackers had exploited a previously unknown flaw in Metabase, a self-hosted business intelligence application it used. Names, addresses, phone numbers, birth dates and prescription information were potentially taken, and the Direwolf ransomware and extortion group claimed responsibility. Letters began going out on October 8. Separately, Healthcare Highways, a Frisco, Texas provider network and health plan administrator, found that attackers had been in its network from July 22 to July 27, 2026, with Social Security numbers, diagnoses, claims and payment information among the data accessed; the Chaos group claimed 235 GB. The lesson for health tech firms is that a reporting tool built to read everything becomes the shortest path to everything when it is exposed.
The Six Points to Take Away
- Analytics tools have broad read access by design. A business intelligence server usually connects to the most complete copy of your data.
- Self-hosted means self-patched. When you run the software yourself, its flaws and updates are your responsibility, even if the vendor writes the fix.
- Zero-days do not wait for patch day. Photon’s attackers used a flaw nobody had a fix for yet, so exposure and access limits had to do the work.
- Extortion groups target data, not uptime. Both incidents involved theft claimed by ransomware and extortion groups.
- Prescription data is sensitive health data. Even without Social Security numbers, prescription details expose conditions and treatments.
- Notification follows weeks of review. Photon finished its review on September 4 and began notifying in October.
The six fixes below focus on reducing what an exposed analytics server can reach and how quickly misuse is noticed.
Why Self-Hosted Analytics Tools Are a Prime Target
Self-hosted analytics tools are a prime target because they sit at the meeting point of convenience and access. Teams install a business intelligence tool so product managers, analysts and executives can query data without asking engineers. To make that work, the tool connects to production databases or a data warehouse with read access to many tables, and it is often reachable from the internet so staff can use it anywhere.
Our team sees this setup frequently at growing health tech companies. The analytics server was stood up quickly during a growth phase, connected with a broad database account, and then left alone because it worked. Nobody owned its updates, its login settings or its network exposure. When a zero-day appears in software like that, the attacker inherits whatever the tool can read, which is usually a great deal.
A fair counterpoint is that no company can patch a flaw that has no patch. That is exactly right, and it is why the defenses here focus on exposure and access rather than speed of updating alone. A zero-day in a tool that only a few people can reach, through a private network, with read access to a limited set of tables, is a much smaller incident.
Why business intelligence tools hold so much data
Business intelligence tools hold so much data because their job is to answer any question someone asks. Analysts want joins across patients, prescriptions, claims and billing, so the tool’s database account is often granted access to all of them.
What self-hosting changes
Self-hosting changes who carries the patch. A software-as-a-service vendor patches its own servers; a self-hosted copy is only patched when your team downloads and applies the update. It also means your network rules decide who can reach the login page.
Why the second incident still matters
Healthcare Highways did not name the entry point, but its data types, including Social Security numbers, procedure codes and payment information, show how much a health plan administrator holds. Five days of access was enough for the attackers to claim a large data theft.
6 Fixes for Self-Hosted Analytics at Health Tech Firms
The six fixes below assume the next flaw will arrive before its patch does.
Fix 1: Take analytics servers off the open internet
Removing public exposure is the first fix. Put the analytics tool behind a VPN, zero-trust access gateway or single sign-on proxy so only authenticated staff on approved devices can reach its login page. Most exploitation of self-hosted tools starts with an internet scan.
Fix 2: Give the tool a narrow, read-only database account
Narrowing database access is the second fix. Connect the analytics tool through an account that can read only the tables and columns it needs, apply a least privilege model, and exclude identifiers like Social Security numbers unless a report truly needs them.
Fix 3: Point analytics at a de-identified copy where possible
Using a de-identified copy is the third fix. Many dashboards work just as well on a reporting database with masked names and dates. If the tool is breached, the stolen data is far less harmful.
Fix 4: Track and patch self-hosted tools like core systems
Patching like a core system is the fourth fix. Add every self-hosted tool to your asset inventory and patch management process, subscribe to its security advisories and assign a named owner. A vulnerability assessment will often find tools nobody listed.
Fix 5: Separate analytics from production networks
Separating networks is the fifth fix. Place analytics servers in their own segment, as described in our guide to network segmentation for ePHI, so a compromised tool cannot be used to reach other systems.
Fix 6: Alert on large queries and exports
Alerting on unusual queries is the sixth fix. Large exports, queries across many tables at odd hours and new outbound connections from the analytics server are signs of data exfiltration. Network security monitoring can flag them in real time.
Questions to ask before installing the next self-hosted tool
Before any team installs another self-hosted tool that touches patient data, ask four questions. Who owns its updates and how will they hear about a security advisory? Will it be reachable from the internet, and if so, why? Which database account will it use, and what can that account read? How will anyone notice if it starts exporting far more data than usual? If nobody can answer all four, the tool is not ready for production, however useful it looks in a demo. Writing the answers down takes an hour and turns a quick install into a system someone is accountable for.
How to Prepare for Data Extortion
Preparing for data extortion means planning for theft rather than only for encryption.
Know what each system holds
Map which systems hold which patient data so you can tell quickly what an attacker could have taken, a core part of any HIPAA risk assessment.
Plan the notification path
Decide who leads the forensic review and how you will meet the HIPAA breach notification rule timelines once the data is known.
Decide your extortion stance in advance
Agree with leadership, counsel and your insurer how you will respond to a demand, because double extortion groups apply pressure fast.
Frequently Asked Questions
What happened in the Photon Health breach?
Attackers exploited a previously unknown flaw in Metabase, a self-hosted business intelligence tool Photon Health used, and potentially took names, addresses, phone numbers, birth dates and prescription information. The Direwolf group claimed responsibility.
What happened at Healthcare Highways?
Unauthorized parties accessed Healthcare Highways’ network from July 22 to July 27, 2026, reaching data including Social Security numbers, diagnoses, claims and payment information. The Chaos group claimed 235 GB of stolen data.
Why are business intelligence tools risky?
Business intelligence tools usually have broad read access to company databases. If one is exposed and compromised, attackers can query most of the data it connects to.
How can we protect a self-hosted analytics tool from a zero-day?
Remove it from the public internet, give it a narrow read-only database account, use de-identified data where possible and alert on unusual queries and exports.
Should health tech firms avoid self-hosted tools?
Not necessarily. Self-hosted tools can be safe when they are patched, isolated and limited in access. The risk comes from treating them as low-priority utilities.
Who Is Behind This Advice
This guidance comes from the healthcare IT and security teams at Mindcore Technologies, who help health technology companies, provider networks and practices secure data platforms, analytics tools and the networks around them. The six fixes above reflect the review we run when a client’s reporting stack touches patient data. Our role is to give your team clear priorities and practical settings, so decisions about your systems stay with you. Mindcore is led by CEO Matt Rosenthal, who focuses on helping healthcare organizations protect patient trust across every tool that reads their data.
Treat the Analytics Server as a Crown-Jewel System
The Photon Health incident shows that a reporting tool can become the shortest path to a health company’s most sensitive data. Take analytics servers off the open internet, give them narrow read-only access, use de-identified copies where you can, patch them like core systems, separate them from production networks and alert on unusual queries. Prepare for data extortion with a clear map of what each system holds and a notification plan ready to use. If you want help reviewing your analytics stack or the access it holds, book a free strategy call and our team will walk through it with you.


