Skip to main content
Return to TrendAI™ Security Blog
Ransomware

BYOD Ransomware: An Identity-First Extortion Campaign Moving Into Hospitals and Universities

In about five weeks, a new ransomware group called BYOD went from small businesses to a trauma center and two major U.S. universities. Here’s what we know and what to hunt for now.

Education Healthcare & Life Sciences Ransomware & extortion General Markets

Key Takeaways

  • BYOD has claimed at least nine organizations in five weeks, and three major university or health organizations in the two weeks from October 26 to November 9.
  • Every account of its access points to identity: Active Directory, Group Policy, a cloud identity tenant, a domain controller, a credential compromise and a remote access trojan on an employee's computer. No CVE has been tied to any BYOD attack.
  • Its attacks have had real-world consequences, including ambulance diversion, more than 30 clinics closed, and campus-wide device shutdowns.
  • BYOD runs a pressure campaign alongside the intrusion: audio messages on infected computers, public leak-site posts, statements framing its attacks as holding organizations accountable, and outreach to individual victims.
  • Its claims are larger than what victims have confirmed. UNM Health Sciences Center, Michigan State and Vanderbilt have each disputed or not confirmed parts of them.
  • Defenders in higher education, healthcare and the public sector should treat directory services and cloud identity as the front line and hunt now, rather than wait for indicators.

Summary

In the five weeks between October 5 and November 9, 2026, a previously unknown ransomware and data-extortion group calling itself BYOD publicly claimed at least nine organizations. It began with small businesses, moved to a national consumer phone service, and then hit the University of New Mexico (UNM) Health Sciences Center , Michigan State University , Vanderbilt University , and the Texas Health and Human Services Commission . At UNM, the attack forced the state's only Level 1 trauma center to divert ambulances. At Vanderbilt, the university told its entire community to shut down university devices.

Taken one at a time, these look like separate incidents. Taken together, they show a single group with a consistent playbook: it goes after identity infrastructure rather than software vulnerabilities, takes data at scale before it encrypts, and uses deliberately public pressure, including audio ransom messages played on victims' computers, to force a response. No vulnerability, malware sample or indicator of compromise has been published for BYOD. This analysis is built from victim statements, public reporting and the group's own claims, and it is explicit about which is which. It explains how the attacks connect, what the reported activity tells us about how BYOD operates, why higher education and healthcare are exposed to it, and what defenders can hunt for and fix now.

Who BYOD is

BYOD is a ransomware and data-extortion group that was first observed in September 2026. The name matches the common IT term for "bring your own device," but the group has not explained it and nothing links it to that term. It runs a data leak site where it lists victims, publishes stolen data when victims do not engage, and posts statements about its attacks. By October 27, six victims were listed and data had already been published for three of them.

The group presents itself in unusual terms. Its leak site describes its activity as "pentesting" meant to expose poor security and hold the organizations that hold personal data accountable. It says no target is off-limits, that it chooses victims after an initial assessment of whether they are a target of interest, and that it focuses on organizations it sees as treating customers as "a number." Victims are directed to negotiate through the Session encrypted messenger. At Vanderbilt, the ransom note was named README_BYOD_31.txt.

Nothing public yet identifies who is behind BYOD, what its encryptor is based on, or whether it operates as a closed group or recruits affiliates.

Timeline

Dates are when the victim appeared on BYOD's leak site or was first reported.

  • October 5: Franklin Empire and Standpointe / Trinite Solutions, two smaller businesses. Leak-site listings only.
  • October 5: Trump Mobile. Data on 3,615 customers is published and reported, and BYOD takes responsibility.
  • October 13: Decorolala, a smaller business. Leak-site listing only.
  • October 26: University of New Mexico (UNM) Health Sciences Center. Ambulances are diverted and clinics close. BYOD lists UNM the same day.
  • November 1: Michigan State University is listed. The university links the claim to a credential compromise it detected in early October.
  • November 3: IHG Hotels and Resorts is listed, followed by a "first and final warning." IHG has not publicly confirmed an incident.
  • November 9: The Texas Health and Human Services Commission is listed. The agency has not publicly confirmed an incident.
  • November 9: Vanderbilt University announces an active cybersecurity incident. BYOD claims the attack the same day.

Incident analysis

UNM Health Sciences Center (October 26)

UNM Health Sciences Center detected a cyberattack on October 26 that affected the entire Health Sciences Center, which includes its hospitals, medical school and research facilities. It diverted ambulance traffic to other hospitals from 12:30 to 2:30 p.m., cancelled appointments and outpatient surgeries at more than 30 clinics, and cancelled classes across its medical, nursing, pharmacy and population health colleges. UNM Hospital is New Mexico's only Level 1 trauma center, and the system handles roughly 900,000 outpatient visits and 100,000 emergency room visits a year, so for two hours the state had no emergency department able to take its most seriously injured patients. Staff were told to unplug desktops and take mobile devices off hospital Wi-Fi, and employees reported messages playing across their computers claiming that patient, student and employee data had been stolen. The FBI is assisting.

BYOD claimed it had compromised 36,000 systems across UNM's main campus and hospital, including its Active Directory forests, its Group Policy infrastructure, its entire Microsoft Entra tenant and two Oracle databases, and said it held student and employee personal data. UNM's own assessment is narrower: it says the attack affected some end-user devices but did not penetrate core systems or servers. Critical systems were back online early the next morning, all facilities were operating normally within days, and UNM says it did not pay a ransom. It warned patients, students and staff not to respond to anyone claiming responsibility and to report any contact to the FBI.

After the disruption at UNM, BYOD posted apologies, said it had not known the affected systems supported patient care, and offered free decryption keys to healthcare facilities, while saying it would keep the stolen data. The offer of keys indicates that the group encrypted systems at UNM.

What it tells defenders: the gap between BYOD's claim (the whole identity estate) and UNM's assessment (some end-user devices) is the central question in every BYOD case. Even on UNM's account, encryption of end-user devices alone was enough to take a trauma center offline for two hours. Fast containment, clear staff instructions and working downtime procedures are what shortened the outage.

Michigan State University (claimed November 1)

BYOD listed Michigan State University on November 1 and claimed personal information on 1.3 million current and former students. Michigan State says it became aware of suspicious activity in early October that appeared to be a potential credential compromise, and on October 1 required every holder of a university account to reset their password to one of at least 15 characters. The university says current evidence does not show a compromise of its central Active Directory and that it has no evidence other data was accessed, changed or stolen. It has not commented on BYOD's specific claims.

What it tells defenders: this is the clearest confirmed link between BYOD and stolen credentials. The leak-site post came about a month after the university's own response. That lag is consistent with a group that steals data first and extorts later, and it means an organization that has already handled a credential incident should still check what those accounts reached before the reset.

Trump Mobile (early October)

In early October, data on 3,615 Trump Mobile customers was published, including names, home addresses, email addresses, phone numbers and order details. People in the dataset confirmed it was accurate, and it included some who had never completed sign-up. BYOD claimed responsibility. A member of the group said they got in by installing a remote access trojan on the computer of an employee of the mobile carrier that runs the service, and the group also claimed access to the website's back end. The leak drew congressional attention.

What it tells defenders: this is the only account, from the group itself, of how BYOD gains initial access, and it points to an endpoint compromise of a third party's employee rather than an exploit. It also shows the group's interest in high-visibility targets: a small dataset from a well-known brand brought national attention.

Vanderbilt University (November 9)

Vanderbilt, which serves about 14,000 students and more than 67,000 staff , announced on November 9 an active cybersecurity incident that could affect university devices, systems and cloud-based services. It told students, faculty and staff to shut down university devices and disconnect from campus Wi-Fi, and warned them to watch for unexpected pop-ups and security warnings, unexpected requests to approve logins or enter passwords, unfamiliar applications, and unusual changes to files and permissions in OneDrive, Box and other cloud storage. Campus computers played an audio message: "You have been compromised by BYOD," followed by a demand to make contact and a threat to bring the estate "to the ground."

BYOD claims it encrypted Vanderbilt's databases, took full control of its domain controller, and stole research data and more than one million Social Security numbers along with names, contact details, birthdays, roles and student and employee records. The ransom note links to BYOD's leak site and its Session contact. Vanderbilt restored campus Wi-Fi the same evening and brought single sign-on, email, its learning platform and video conferencing back in phases over the following days, while other systems stayed restricted. It has not confirmed BYOD's involvement, whether systems were encrypted, or what data was taken.

What it tells defenders: Vanderbilt's warnings are a useful map of what it was seeing or expected to see. Unexpected login approval requests point to abuse of authentication; permission changes in cloud storage point to access to collaboration platforms through compromised identities. Both sit outside the traditional endpoint and server view of a ransomware attack.

Other victims

IHG Hotels and Resorts was listed on November 4, followed by a "first and final warning" post on November 5. The Texas Health and Human Services Commission was listed on November 9. Neither has publicly confirmed an incident. Three smaller businesses listed in October appear only on the leak site.

How BYOD operates: the reported attack chain

No full intrusion has been publicly analyzed, so the chain below is assembled from the five victims with public detail. Each step is marked as confirmed by a victim, reported, or claimed by BYOD. It describes what the group appears to do, not a verified sequence for any single incident.

Initial access

Two routes are on record. At Michigan State, the university confirms a probable credential compromise (confirmed). At Trump Mobile, a BYOD member claims a remote access trojan on an employee's computer at a third-party carrier (claimed). No exploited vulnerability has been reported in any case. Both routes put the attacker inside with a legitimate identity, which helps explain why there has been no patch to point to.

Privilege and identity takeover

BYOD's claims consistently name the systems that control identity: Active Directory forests and Group Policy at UNM, the Microsoft Entra tenant at UNM, and the domain controller at Vanderbilt (claimed). Control of Group Policy or a domain controller lets an attacker push settings, scripts and scheduled tasks to every domain-joined machine at once, which is how many ransomware operators deploy an encryptor across an organization in a single step. Control of a cloud identity tenant extends the same reach to email, file sharing and other cloud applications. Vanderbilt's warning about unexpected login approval requests and permission changes in OneDrive and Box is consistent with activity at that cloud identity layer (reported).

Collection and exfiltration

BYOD claims two Oracle databases at UNM, research data at Vanderbilt, and personal records numbering in the millions at Michigan State and Vanderbilt (claimed). The Trump Mobile leak shows it does publish what it takes (confirmed by people in the data). How the data leaves the network has not been reported.

Impact

Encryption is confirmed indirectly at UNM by BYOD's offer of decryption keys, and claimed for databases at Vanderbilt. Both institutions ordered devices shut down to contain the spread (confirmed). The audio messages at UNM and Vanderbilt (reported by staff and students and captured on video) mean the attackers could run code that played sound on many machines at once. That fits with, though it does not prove, the broad, directory-level access BYOD claims.

Extortion

BYOD lists victims on its leak site, posts warnings such as IHG's "first and final warning," publishes data, and makes offers aimed at individuals as well as the organization (confirmed by the leak-site posts as reported). Negotiation runs through Session.

Mapping to MITRE ATT&CK

The table maps reported and claimed behaviors to ATT&CK techniques for hunting and coverage review. It reflects the public record only and will change as technical analysis becomes available.

Stage Behavior Where Status ATT&CK
Initial access Use of stolen credentials Michigan State Confirmed by victim T1078 Valid Accounts
Initial access Remote access trojan on a third-party employee's computer Trump Mobile Claimed by BYOD Technique not public
Privilege escalation Control of domain controller and domain accounts Vanderbilt, UNM Claimed by BYOD T1078.002 Domain Accounts
Defense evasion / deployment Control of Group Policy infrastructure UNM Claimed by BYOD T1484.001 Group Policy Modification
Persistence Control of cloud identity tenant UNM Claimed by BYOD T1078.004 Cloud Accounts
Credential access Unexpected login approval requests reported to users Vanderbilt Reported by victim T1621 Multi-Factor Authentication Request Generation (possible)
Collection Theft from databases and file stores UNM, Vanderbilt Claimed by BYOD T1213 Data from Information Repositories
Exfiltration Bulk theft of personal and research data All named victims Claimed; published for Trump Mobile Method not public
Impact Encryption of systems and databases UNM, Vanderbilt UNM implied by key offer; Vanderbilt claimed T1486 Data Encrypted for Impact
Impact Audio ransom message played on infected computers UNM, Vanderbilt Reported by victims' staff and students T1491.001 Internal Defacement (closest match)
Table 1. MITRE ATT&CK matrix of observed BYOD ransomware incidents

The pressure model

BYOD's most distinctive feature is not technical. It is the way the group turns an intrusion into a public event and then spreads the pressure from the institution to the people whose data it holds.

Audio ransom messages are the clearest example. A ransom note in a folder may be seen by a few administrators; a voice announcing the compromise from computers across a hospital or campus is heard by staff, patients and students within minutes, recorded on phones and posted online. The incident becomes public before the institution has finished its first assessment, and its communications team is working against a story the attackers have already started.

The group then reaches past the institution. After UNM, it offered deals aimed at individuals and kept the data it said it would not use against patients. UNM's warning not to engage with anyone claiming responsibility shows the institution expected direct contact. For universities and hospitals, that means the extortion target is not only the organization but also tens of thousands of students, patients and employees, each of whom can be frightened, scammed or pressured.

Finally, BYOD wraps its attacks in a justification. Calling its activity "pentesting," claiming to hold organizations accountable, and apologizing for the hospital disruption are reputation management. They aim to make paying, or engaging, seem more reasonable, and to make the victim's security the story instead of the crime.

TrendAI™ Research has documented elements of this model in other groups. Our Ransomware Spotlight on DragonForce noted that the group published an audio recording of a victim in 2024 and later launched a "data analysis service" to build tailored extortion material for affiliates. Our spotlight on Agenda described leak-site countdown timers and staged data releases used to escalate pressure. BYOD puts these pressure tactics at the center of its operation and applies them to targets where disruption reaches patients and students directly.

How BYOD compares with groups we track

Several of BYOD's reported behaviors match patterns TrendAI Research has documented in established ransomware operations, which helps defenders anticipate what is not yet public.

  • Stolen credentials as the way in. Our Agenda spotlight describes the group logging in with stolen credentials and using access purchased from initial access brokers before elevating privileges. Michigan State's credential compromise fits the same pattern.
  • Directory-level reach. Agenda uses PowerShell and Active Directory commands to enumerate and spread across domain environments. BYOD's claims about Active Directory, Group Policy and the domain controller describe the same goal.
  • Fast, interchangeable payloads. Our DragonForce spotlight shows how a new group can launch quickly on leaked LockBit and Conti builders and switch payloads to stay ahead of defenses. Whether BYOD's encryptor is new or built on leaked code is not yet known, and it is one of the first questions a sample would answer.
  • Data theft as the main lever. Both groups combine encryption with data theft and leak-site pressure. BYOD's claims emphasize stolen personal data above encryption, and its offer of free keys at UNM shows it is willing to give up the decryption leverage while keeping the data.

What is different is the target profile and the speed. In five weeks BYOD went from small businesses to a trauma center and two major US universities, with a heavy concentration in education and healthcare.

Why higher education and healthcare are exposed

Universities and academic medical centers are among the hardest environments to protect at the identity layer, and BYOD's claims target exactly that layer.

  • Scale and turnover. A large university manages tens of thousands of accounts for students, faculty, staff, researchers, clinicians and affiliates, with thousands joining and leaving every term. Stale accounts, inconsistent multifactor enrollment and excess permissions build up.
  • Hybrid identity. Most run on-premises Active Directory synchronized with a cloud identity tenant, so control of one side can lead to the other. BYOD's claim at UNM names both.
  • Decentralized IT. Departments, labs and clinical units often run their own systems and administrators, which multiplies privileged accounts and makes changes harder to spot.
  • Open by design. Academic networks are built for collaboration and visiting users, and research data, student records and patient records often share infrastructure.
  • High-value data and high-stakes uptime. Universities hold Social Security numbers, financial aid and research data; academic medical centers add patient records and life-safety systems. Downtime is immediately visible and politically costly, which raises the pressure to settle.

What the trend is

BYOD shows how quickly a new group can move from obscurity to attacks with life-safety consequences. Each step in its five-week path brought larger datasets, more users and more public attention: small businesses, a consumer brand, a university hospital system, two major universities and a state health agency. In November it claimed a new victim every few days.

The defining features of this campaign are that it is identity-first, that it does not depend on a vulnerability, and that it is built for public pressure. Together they mean the usual triggers for urgent action, a CVE, a patch or a published set of indicators, are absent. Organizations that wait for those signals will have nothing to act on until they are already a victim.

Healthcare and higher education have long been heavily targeted by ransomware. BYOD's contribution to that picture is speed, a sharp focus on the identity layer, and a deliberate strategy of making each attack visible to everyone it affects. More education, healthcare and public sector organizations should expect to be on its list.

What we don't know yet

  • How BYOD gets in at most victims. The only accounts are a credential compromise at Michigan State and the group's own claim of a trojan on an employee's computer at Trump Mobile.
  • What its encryptor is, whether it is new or built on leaked code, and which platforms it can encrypt.
  • How the audio messages are delivered, and how the group exfiltrates data.
  • How much of its claimed access and theft is real. UNM, Michigan State and Vanderbilt have each disputed parts of the claims or not confirmed them.
  • Who is behind it, whether it works with affiliates, and whether it is linked to other groups.

Detection and hunting guidance

Without published indicators, defenders should hunt for the behaviors the reporting describes. The following are starting points across identity, endpoint, cloud and network.

Directory and Group Policy

  • New or changed membership in privileged groups such as Domain Admins and Enterprise Admins (Windows Security events 4728, 4732 and 4756), and new accounts created outside normal provisioning (4720).
  • Changes to Group Policy objects, especially new scheduled tasks, startup scripts or software installation settings (directory service changes, event 5136, and changes in SYSVOL).
  • Administrator sign-ins to domain controllers from workstations or at unusual times, and use of remote execution tools such as PsExec or remote PowerShell against many hosts in a short window.

Cloud identity

  • New role assignments for Global Administrator and other privileged roles, new credentials added to applications or service principals, and changes to conditional access or authentication policies.
  • Bursts of multifactor authentication prompts for the same user, approvals from unfamiliar locations or devices, and new authentication methods registered to existing accounts.
  • Mass permission changes, sharing-link creation or downloads in OneDrive, SharePoint, Box and similar platforms.

Endpoints and servers

  • Remote access tools or unknown remote access software on employee endpoints, especially for staff with access to sensitive data and for third-party and contractor devices.
  • Scripts or processes playing audio or using text-to-speech on many machines at once, which would be highly unusual in a hospital or campus environment.
  • Mass file renaming or modification, deletion of shadow copies and backups, and services stopped across many hosts.

Data movement

  • Large exports from databases such as Oracle and from student, HR and research systems, and creation of large archives on servers.
  • Unusual outbound transfers to cloud storage or file-sharing services, and large uploads from servers that do not normally send data out.

What to do

Protect identity first

  • Review privileged accounts in Active Directory and cloud identity tenants. Remove stale and excess admin rights, separate administrator accounts from everyday accounts, and restrict where administrators can sign in.
  • Require phishing-resistant multifactor authentication for remote access, administrators and cloud identity, and enable number matching or equivalent protection against approval fatigue.
  • Protect the systems that synchronize on-premises and cloud identity as carefully as domain controllers.
  • Treat a confirmed credential compromise as a possible precursor. Reset the passwords, then review what those accounts touched in the weeks before.

Find the attack before encryption

  • Use the hunting guidance above across identity, endpoint, cloud and network together. In the attacks described here, the warning signs span all four.
  • Inventory and restrict remote access software, and review access held by contractors and partners.
  • Alert on bulk data access and outbound transfers from systems that hold student, patient and research data.

Be ready to recover and communicate

  • Keep offline, tested backups of directory services and critical databases, and rehearse restoring identity infrastructure from scratch.
  • Prepare a short, clear instruction for users on what to shut down, what to watch for and where to report it. UNM and Vanderbilt both relied on this to contain the damage.
  • For hospitals, maintain diversion and downtime procedures that do not depend on the network.
  • Warn students, patients and staff that criminals may contact them directly about stolen data, and tell them not to respond, click links or pay, and to report the contact.

How TrendAI™ can help

TrendAI™ Research is confirming whether existing TrendAI™ detections cover BYOD's encryptor and tooling. No BYOD samples or indicators have been published, so this report does not yet name a BYOD-specific detection; the specific detections will be added here once confirmed.

BYOD's attacks cross identity, endpoint, cloud and network, and its warning signs only make sense when they are seen together: an unusual admin sign-in, a Group Policy change, a burst of login prompts, a large export. TrendAI Vision One™ correlates activity across endpoint, network, cloud, email and identity sensors in a single platform, so that these signals can be investigated as one attack, before encryption.

For organizations that need help now, TrendAI Vision One™ Services – Incident Response provides expert-led investigation, scoping, containment and recovery when an incident goes beyond internal capacity, including digital forensics, root-cause and attacker-behavior analysis, and secure communication outside a compromised network. TrendAI Vision One™ Services – Managed Detection and Response (MDR) adds around-the-clock monitoring, threat hunting, investigation and response for teams that cannot watch identity, endpoint and cloud activity at all hours.

TrendAI™ Research continues to track BYOD and will update this analysis as technical details emerge. Our Ransomware Spotlight series, including Agenda and DragonForce , covers related techniques in more depth.