Skip to content

Response, New York. Noah Park: Check Point Patches Actively Exploited Management Server Path Traversal. Detection, Dallas. Ava Okello: Report Finds Nearly Half of Deployed Detections Need Attention Before SOCs Can Trust Them.

SOCtember

Always First. Fast SOC News.

Screenshot of the TrendAI Security Blog post titled BYOD Ransomware: An Identity-First Extortion Campaign Moving Into Hospitals and Universities, dated November 18, 2026, with the opening of its Key Takeaways list.
Image: TrendAI (fair use). The opening of TrendAI Research's Nov. 18 analysis of the BYOD ransomware group.

Detection

TrendAI Says the New BYOD Ransomware Crew Went From Small Businesses to a Trauma Center and Two Major Universities in Five Weeks by Going After Identity, Not Software Flaws

No vulnerability, malware sample or indicator has been published for BYOD, which has claimed at least nine victims since Oct. 5 and played spoken ransom demands on hospital and campus computers. TrendAI's hunting guidance starts with privileged group changes, Group Policy edits and bursts of login approval prompts.

Ava Okello, DetectionLondon6 min read

LONDON - A ransomware group that calls itself BYOD has moved in about five weeks from listing small businesses on its leak site to disrupting a Level 1 trauma center and two major American universities, and every public account of how it operates points to stolen identities rather than an exploited flaw, according to an analysis published Nov. 18 by TrendAI Research.

The report is unusual in what it lacks. "No vulnerability, malware sample or indicator of compromise has been published for BYOD," the researchers wrote, so their account is assembled from victim statements, public reporting and the group's own claims, with each step labeled as confirmed by a victim, reported, or claimed by BYOD. "No CVE has been tied to any BYOD attack," the report says.

For security operations teams, that is the point. "The usual triggers for urgent action, a CVE, a patch or a published set of indicators, are absent," TrendAI wrote. "Organizations that wait for those signals will have nothing to act on until they are already a victim."

Five weeks, nine claimed victims

TrendAI counts at least nine organizations claimed between Oct. 5 and Nov. 9. The list opens with smaller businesses and Trump Mobile, whose data on 3,615 customers was published in early October. On Oct. 26, the University of New Mexico Health Sciences Center detected an attack that, by TrendAI's account, led it to divert ambulances from 12:30 to 2:30 p.m., cancel appointments and outpatient surgeries at more than 30 clinics and cancel classes. UNM Hospital is the state's only Level 1 trauma center.

Michigan State University was listed on Nov. 1, IHG Hotels and Resorts on Nov. 3 and the Texas Health and Human Services Commission on Nov. 9, the same day Vanderbilt University announced an active cybersecurity incident and told students, faculty and staff to shut down university devices and disconnect from campus Wi-Fi. IHG and the Texas agency have not publicly confirmed incidents, the report notes.

Screenshot of the Timeline section of TrendAI's BYOD report, listing leak-site claims from October 5 to November 9, 2026, including Trump Mobile, UNM Health Sciences Center, Michigan State University, IHG Hotels and Resorts, the Texas Health and Human Services Commission and Vanderbilt University.
Image: TrendAI (fair use).The report's timeline of BYOD's claimed victims.

The claims and the confirmations diverge sharply. BYOD said it compromised 36,000 systems at UNM, including its Active Directory forests, Group Policy infrastructure, its entire Microsoft Entra tenant and two Oracle databases. UNM's own assessment, as summarized by TrendAI, is that the attack affected some end-user devices but did not penetrate core systems or servers. At Vanderbilt, BYOD claims it took full control of the domain controller and stole more than one million Social Security numbers; the university has not confirmed the group's involvement, encryption or data theft.

Ransom demands spoken aloud

The group's most visible tactic is audio. Staff at UNM reported messages playing across their computers claiming that patient, student and employee data had been stolen. At Vanderbilt, TrendAI wrote, campus computers played the message "You have been compromised by BYOD," followed by a demand to make contact. The ransom note there was named README_BYOD_31.txt.

TrendAI is careful about what that proves. The audio means "the attackers could run code that played sound on many machines at once," the report says, which "fits with, though it does not prove, the broad, directory-level access BYOD claims." Control of Group Policy or a domain controller, it notes, would let an attacker push scripts and scheduled tasks to every domain-joined machine in a single step.

The only confirmed link to stolen credentials comes from Michigan State, which detected a likely credential compromise in early October and required every account holder to reset to a password of at least 15 characters. BYOD listed the university about a month later. The only first-hand account of initial access comes from the group itself, which said it reached Trump Mobile through a remote access trojan on the computer of an employee at the third-party carrier that runs the service.

Screenshot of Table 1 in TrendAI's BYOD report, a MITRE ATT&CK matrix mapping reported and claimed behaviors such as stolen credentials, Group Policy control, unexpected login approval requests, data theft, encryption and audio ransom messages to techniques including T1078, T1484.001, T1621, T1486 and T1491.001.
Image: TrendAI (fair use).TrendAI's ATT&CK mapping, with each behavior marked as confirmed, reported or claimed.

What TrendAI says to hunt

With no indicators to load, the report gives behaviors. In the directory, it points to new or changed membership in privileged groups such as Domain Admins and Enterprise Admins, which surface as Windows Security events 4728, 4732 and 4756, and accounts created outside normal provisioning, event 4720. It flags changes to Group Policy objects, "especially new scheduled tasks, startup scripts or software installation settings," visible in directory service change event 5136 and in SYSVOL, along with administrator sign-ins to domain controllers from workstations or at unusual times and PsExec or remote PowerShell fanned out to many hosts.

In cloud identity, it lists new Global Administrator and other privileged role assignments, new credentials added to applications or service principals, changes to conditional access, bursts of multifactor prompts for one user, new authentication methods on existing accounts and mass sharing or downloads in OneDrive, SharePoint and Box. On endpoints, it calls out unknown remote access software and "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," plus mass file renaming and shadow copy deletion. For data movement, it names large exports from Oracle, student, HR and research systems and unusual uploads from servers that do not normally send data out.

Screenshot of the Detection and hunting guidance section of TrendAI's BYOD report, listing directory and Group Policy, cloud identity, and endpoint and server behaviors to hunt, including Windows Security events 4728, 4732, 4756, 4720 and 5136.
Image: TrendAI (fair use).The report's hunting starting points across directory, cloud identity and endpoints.

What it means for the SOC

Much of that telemetry is already collected in a typical enterprise; the gap the report exposes is correlation. A Group Policy edit, a run of approval prompts and a large OneDrive export can each look routine to the team that owns that log source. TrendAI's argument is that, for this group, they are the attack, and they appear before encryption.

The Michigan State sequence adds a practical rule for responders. "Treat a confirmed credential compromise as a possible precursor," the report says. "Reset the passwords, then review what those accounts touched in the weeks before." A password reset that closed an incident in early October did not, on the group's timeline, close the extortion that followed.

TrendAI also recommends phishing-resistant multifactor authentication for remote access, administrators and cloud identity, protecting the systems that synchronize on-premises and cloud identity "as carefully as domain controllers," and keeping offline, tested backups of directory services. It says it will update the analysis as samples and technical detail emerge.

About the Author

Ava Okello

Ava Okello covers detection engineering, EDR telemetry, and SOC hunting for SOCtember from London.

Related stories

Detection desk