Ransomware family
DragonForce Ransomware Decryption & Data Recovery
- Active
- Critical
- No public decryptor
DragonForce is one of the most active ransomware cartels today. Since 2025 it has offered white-label encryptors and infrastructure to affiliates, hits virtualisation estates hard, and became widely known through the chain of UK retail attacks. No public decryptor exists.
- First seen
- 2023-08
- File extensions
- .dragonforce_encrypted .df_win .[9位随机字符]
- Ransom notes
- readme.txt
- Affected platforms
- Windows / Linux / VMware ESXi / NAS storage
Family profile
- File extensions
- .dragonforce_encrypted
- .df_win
- .[9位随机字符]
- Ransom notes
- readme.txt
- README.txt
- Contact Us.txt
- Contact patterns
- Tor (.onion) negotiation portal + per-victim token
- DragonForce / RansomBay leak sites on Tor
- Affiliate-branded contact details (cartel model)
- Nuisance calls and voice social engineering (with Scattered Spider-style access brokers)
- Aliases / versions
- DragonForce Ransomware Cartel、RansomBay(白牌服务)、DragonForce ESXi Locker
- First seen
- 2023-08
- Status
- Active
- Threat level
- Critical
- Affected platforms
- Windows
- Linux
- VMware ESXi
- NAS storage
- Tags
- Active
- Ransomware-as-a-Service
- Double extortion
- Targets virtualization
- Exploits vulnerabilities
- Supply chain
- Leak-site regular
No free public DragonForce decryptor exists. The encryptor is built from the leaked LockBit 3.0 builder and modified Conti v3 source, wrapping per-file symmetric keys with asymmetric cryptography, and no exploitable implementation flaw has been published. Tools advertised online as DragonForce decryptors are untrustworthy and some carry secondary ransomware or infostealer payloads. Recovery must run through backups and snapshots, structural repair of virtual disks and databases, unencrypted duplicates and low-level carving, driven by an assessment of the actual samples.
Latest activity
Symantec found DragonForce hiding C2 inside Microsoft Teams TURN relays via a Go backdoor, Backdoor.Turn, undetected for about two months at a US services firm; entry via MSSQL, EDR killed with vulnerable drivers.
SourcesDragonForce patched encryption flaws once aired publicly and linked to Akira, and abuses vulnerable drivers (truesight.sys, rentdrv2.sys) to disable security tools. Decryption via implementation bugs stays closed.
SourcesReliaQuest reported DragonForce allying with LockBit and Qilin to share tooling, infrastructure and affiliates. Note branding maps even less reliably to the actual encryptor, so attribution must rest on sample analysis.
Sources
Overview
DragonForce first appeared in August 2023, initially generating payloads straight from the leaked LockBit 3.0 builder. It opened an affiliate programme in June 2024, shipped its own encryptor built on modified Conti v3 source in July 2024, and added a Linux variant tuned for VMware ESXi in November 2024.
The turning point came in March 2025, when DragonForce rebranded itself as a ransomware cartel and launched RansomBay, a white-label service letting affiliates run their own brand while DragonForce supplies the encryptor, negotiation portal, leak-site hosting and technical support for roughly a 20% cut. The practical consequence is that ransom-note branding varies wildly across incidents while the underlying encryptor and infrastructure stay the same, which genuinely complicates family attribution.
In spring 2025 DragonForce entered mainstream awareness through the chain of UK retail attacks: Marks & Spencer, Co-op and Harrods were hit within weeks. Public analysis attributes initial access to English-speaking crews in the Scattered Spider mould - specialists in voice social engineering, help-desk impersonation and MFA fatigue - with DragonForce handling deployment and extortion. That pairing of social-engineering access with a mature encryptor sidesteps defences built primarily around vulnerability scanning.
Activity has not slowed: through the first half of 2026 DragonForce kept posting roughly 28 to 30 victims a month, passing roughly 650 named organisations across 65 countries, concentrated in professional services, manufacturing and technology. No public report confirms a targeted campaign against mainland China organisations, but its preferred entry points - edge vulnerabilities of the Ivanti and Log4j class, compromised RMM platforms enabling supply-chain style spread, and exposed ESXi - are just as common locally, and Chinese branches of multinationals should treat this as in scope.
How to identify it
Extension and naming. The in-house encryptor replaces the entire original filename with a random Base32 string and appends .dragonforce_encrypted, so contract.pdf becomes something like 3krbgdxb.dragonforce_encrypted. Original names cannot be recovered from the encrypted file itself, which materially complicates putting restored data back where it belongs. The .df_win extension has also been observed, while early LockBit 3.0-builder payloads used random-character extensions.
Ransom note. readme.txt (Contact Us.txt in some samples) is dropped on the desktop and in each encrypted directory, and frequently also into web roots such as inetpub and wwwroot so it is visible externally. It supplies a Tor portal and a victim-specific token, with no email address.
Attribution difficulty created by the cartel. Under RansomBay white-labelling, the note may carry an entirely unfamiliar brand and point to a different leak site, while the encryptor behaviour, file trailer structure and ESXi-side command patterns remain identical. Identification therefore depends on sample analysis rather than the brand name.
Other indicators. Mimikatz, PsExec, OpenVAS scanning traces, unusual SMB traffic, and tampered registry keys and security software. In ESXi cases, VM files under /vmfs/volumes are encrypted while the host still boots - the encryptor carries whitelists specifically to preserve host bootability.
Infection vectors
DragonForce's affiliate ecosystem is large and access methods vary, but public analysis keeps surfacing the same categories:
- Edge device exploitation. Ivanti Connect Secure flaws (CVE-2023-46805, CVE-2024-21887, CVE-2024-21893) and Apache Log4j2 among others, used to take a DMZ foothold directly.
- Social-engineering access. Working with Scattered Spider-style crews that call the IT service desk impersonating staff to force password resets, run MFA fatigue campaigns, or use SIM swapping to obtain legitimate accounts. Such intrusions look entirely legitimate in logs and were central to the UK retail chain of incidents.
- RMM platforms. Exploiting flaws in remote management products such as SimpleHelp to enter a service provider environment and then push payloads to downstream customers through the management channel - a supply-chain style spread that is especially dangerous for MSPs and shared group IT platforms.
- Stolen credentials and VPN. Accounts sourced from infostealer logs and dark-web markets.
- Lateral movement and escalation. Mimikatz credential dumping, PsExec, RDP and domain admin takeover, prioritising domain controllers and the virtualisation management plane.
- Exfiltration completed before encryption, feeding double extortion and harassment calls.
From a Chinese perspective the last two deserve the most attention: group enterprises and manufacturers commonly run a shared IT services platform with a broad array of remote maintenance tools, and once that management channel falls, spread is far faster than host-by-host intrusion.
Encryption behavior
Algorithms. The in-house encryptor uses ChaCha8 for content and RSA-4096 to protect each file's symmetric key; early LockBit 3.0-builder payloads inherit LockBit's scheme. Both are designed so that without the operator's private key there is nothing to reverse.
Speed optimisation. Large files are encrypted in blocks with skipping (block size and stride are visible in configuration parameters) so a whole estate can be processed inside a short window. For recovery this means virtual disks and database files may retain long untouched stretches, the main basis for structural repair - though how usable that is must be measured per sample.
Virtualisation. The Linux/ESXi variant encrypts VM files directly under /vmfs/volumes while maintaining whitelists of system directories, extensions and filenames so the host still boots. Keeping the host alive while the business stops is deliberate: it preserves the negotiation channel.
Recovery inhibition. Volume shadow copies deleted, boot-time recovery options disabled, database and backup services stopped, and reachable backup storage and network shares encrypted where possible.
Filename randomisation. Original names are replaced entirely, so restoring directory structure later relies on backup catalogues, database-internal metadata or content-based identification.
Beyond double extortion. The DragonForce ecosystem also sells harassment calling against victims and data analysis services to sharpen extortion leverage.
Assess before you act
Recoverability assessment
No public decryptor exists for DragonForce, and decryption is not cryptographically feasible. We do not pay ransoms and do not negotiate; our work is verifiable technical recovery.
1) Backups, snapshots and shadow copies (highest priority). DragonForce actively destroys reachable backups, but what actually gets destroyed is usually only the online-reachable copy. Sweep systematically for offline media and offsite copies, immutable storage on the backup platform, storage array and NAS volume snapshots, hypervisor snapshots and orphaned snapshot files, cloud version history, and backup servers kept outside domain authentication. In practice this path delivers the highest recovery ratio.
2) Structural repair of VM and database files (depends on the encryption pattern). Because large files are encrypted in blocks, vmdk and vhdx images often retain long intact stretches of guest filesystem metadata and data, so repairing partitions and metadata allows mounting and extraction of inner files. SQL Server, Oracle and MySQL data files support page-level extraction and logical rebuilds with transaction log roll-forward. Yields drop sharply when critical structures are hit, so sample testing precedes any estimate.
3) Unencrypted duplicates and log replay. File-server recycle bins and previous versions, endpoint caches, BI and reporting staging databases, ERP and MES archive exports, mail server copies, and counterparty statements and contract copies can all support reconstruction of critical records.
4) Low-level carving. Where the encryptor wrote new files and deleted originals, source content may survive in unallocated space and can be carved. In ESXi cases this requires never re-initialising the datastore and never creating new VMs on the original LUN.
5) Filename reconstruction. Because original names are randomised, recovered content still has to be put back in place. Backup catalogues, database-internal table names, embedded document metadata and content fingerprints allow directory structure to be rebuilt at scale. This workload is routinely underestimated and belongs in the project plan.
For DragonForce data recovery we provide a measured assessment and a bounded scope. We do not claim "100% decryption", and no technique guarantees full recovery.
Our response plan
Hit by DragonForce ransomware? What to do
Containment and forensic preservation
Cut ESXi hosts, domain controllers and backup servers off from production networks and storage, revoke suspicious sessions and VPN tokens, and suspend push capability on RMM and remote maintenance platforms to prevent redeployment through the management channel. Do not reboot and do not re-initialise datastores. Take read-only images or storage snapshots of datastores and critical volumes, and export ESXi, vCenter, Active Directory, VPN, RMM and service-desk ticket logs - help-desk records are decisive for spotting social-engineered access. Retain three to five encrypted files and readme.txt.
Family identification and encryption analysis
Under a cartel model the note's branding proves nothing. Confirm a DragonForce-lineage payload from the encrypted-file trailer structure, the filename randomisation scheme, ESXi-side command patterns and binary fingerprints, and distinguish the in-house build from early LockBit 3.0-builder payloads. Measure block size and skip stride, and map where intact data survives inside vmdk and database files as input to the repair plan.
Recoverability assessment and plan sign-off
Inventory the real usability of backups and snapshots - not job status but actual mount-and-verify tests - run sample repairs against core VMs and databases, and assess the route and effort for filename reconstruction. Deliver a written assessment stating, per business system, whether recovery runs through backup rollback, structural repair or carving, with expected ranges, timelines and sequence, plus an initial conclusion on exfiltration scope to support compliance decisions.
Recovery execution
Work in a clean environment on images, with original datastores read-only. For ESXi, repair virtual disk structures and mount them to extract data, landing results on newly built clean storage before rebuilding VMs. Databases go through page-level extraction plus transaction log roll-forward. Extracted filesystem data then needs filename and directory reconstruction. Sequence by business impact: identity and network infrastructure, then core production and finance systems, then collaboration and archives. Run integrity and business verification after each batch.
Attribution, hardening and handover
Reconstruct the full kill chain with three entry types under particular scrutiny: unpatched edge appliances, impersonated password or MFA resets at the service desk, and the RMM platform being used as a delivery channel. Hardening: patch and baseline edge devices; introduce identity verification at the help desk (callback, secondary verification); isolate and audit RMM accounts; segment the ESXi management network and enable lockdown mode; reset credentials domain-wide with strong MFA; rebuild backups to 3-2-1 with immutable copies and independent credentials. Close with an incident report, an exfiltration scope conclusion and a handover checklist.
Risk warning
What not to do
- Do not re-initialise the ESXi datastore, create new VMs on the original LUN, or expand or rebuild storage pools - doing so eliminates both structural repair of virtual disks and carving in one move.
- Do not download tools claiming to decrypt DragonForce; no working decryptor exists publicly, and such programs are commonly secondary ransomware or infostealers.
- Do not judge the family from the brand on the ransom note. Under white-label cartel operations an unfamiliar brand may be running the same DragonForce encryptor, and a misattribution leads straight to the wrong recovery plan.
- Do not rush to restore push capability on RMM or remote maintenance platforms, and do not reconnect with the original management credentials - attackers routinely keep that channel for a second deployment.
- Do not reattach backup media or the backup server to an uncleaned network; DragonForce actively searches for and encrypts reachable backup storage.
- Do not issue external statements before the exfiltration scope is established, and do not open private negotiations; payment does not undo the fact that data has already left.
Emergency response
Data already encrypted? Stop and let an engineer look first
We do not pay ransoms and we do not negotiate with attackers. Engineers run a free assessment first, then propose a recovery plan and a firm quote.
Related scenarios
ESXi / Hyper-V Virtualization Encrypted by Ransomware
Hypervisor-level encryption causes the widest blast radius of any ransomware event: dozens of production VMs go dark within an hour or two. This page covers what Linux ESXi encryptors actually do — shut down guests, encrypt vmdk, delete snapshots — the recovery value of flat disk files, and how Hyper-V and Proxmox cases differ.
Domain Controller Compromise and Estate-Wide Encryption
A compromised domain controller hands the attacker a legitimate administrator identity, allowing an encryptor to be pushed to every host at once through Group Policy or remote execution. This page covers how such incidents present, the correct order for Active Directory recovery, and how to decide between cleanup and full rebuild.
Backups Deleted or Destroyed
Modern ransomware follows a fixed sequence: destroy the backups, then encrypt the data — deleting shadow copies, encrypting repositories, disabling jobs, and exploiting backup software flaws to steal credentials. This page covers what can still be inventoried once backups fail, why replication propagates encrypted files off-site, and what offline and immutable copies are really worth.
File Servers and NAS Encrypted by Ransomware
When shared folders on a file server or NAS are encrypted, drawings, contracts, archives, quotations and design sources all become unusable at once — and mapped drives spread the impact to every endpoint. This page covers how to gauge spread, what shadow copies and snapshots realistically offer, and how to sequence recovery by business value.
Related industries
Retail and E-commerce Ransomware Response
In retail and e-commerce, ransomware translates directly into an inability to sell: order systems, membership, POS and warehouse fulfilment stop together and losses accrue by the hour. This page covers the sector's attack patterns, a recovery order built around the order-to-fulfilment chain, and handling of member data exposure.
Manufacturing Ransomware Response and Recovery
Ransomware in manufacturing hits information systems and production cadence at the same time: with ERP down there are no orders, with MES down there is no schedule, and an encrypted drawing library takes the process documentation for an entire product line with it. This page covers the asset profile, recovery priorities and targeted defences.
Construction and Real Estate Ransomware Response
In construction and real estate the core assets are drawings, models and project documentation — typically scattered across project-site NAS units, shared drives and personal computers with no unified backup. This page covers the sector's threat profile, recovery of drawings and BIM models, and defences for dispersed, multi-project environments.
Similar families
- Some versions decryptable
LockBit
LockBit is one of the largest ransomware-as-a-service operations in the world. Despite the 2024 law-enforcement takedown it returned as LockBit 5.0, with working Windows, Linux and VMware ESXi payloads, and it remains one of the most frequently seen families in China.
- No public decryptor
RansomHub
RansomHub launched in February 2024 as a rebrand of Knight/Cyclops and rapidly absorbed affiliates from ALPHV and LockBit with a 90% revenue share, accumulating hundreds of victims within a year. Its infrastructure went offline in early April 2025 and the operation has been dormant since, with affiliates largely migrating to Qilin and DragonForce.
- Some versions decryptable
Akira
Akira is a ransomware-as-a-service operation that emerged in March 2023, breaking in through VPNs without MFA and edge-device flaws, then encrypting Windows estates and VMware ESXi clusters under double extortion. CISA's November 2025 advisory update calls it an imminent threat to critical infrastructure.
FAQ
DragonForce Frequently asked questions
Is there a decryptor for DragonForce-encrypted files?
No. The DragonForce encryptor derives from the leaked LockBit 3.0 builder and Conti v3 source, using ChaCha8 for content and RSA-4096 for key protection, with no published implementation flaw and no trustworthy free tool. Anything online advertised as a DragonForce decryptor should be treated as malware. Real recovery runs through backups and snapshots, structural repair of virtual disks and databases, unencrypted duplicates and low-level carving.
The note does not say DragonForce - why do you attribute it to DragonForce?
Because since March 2025 DragonForce has operated as a cartel, offering white-label service through RansomBay: affiliates run their own brand and leak site while the encryptor and infrastructure remain DragonForce's. The brand therefore carries no identification value; attribution comes from the encrypted-file trailer structure, the filename randomisation scheme, ESXi-side command patterns and binary fingerprints. Getting this right determines the recovery plan, which is why we insist on sample analysis first.
All filenames became random strings - can recovered data be mapped back?
Yes, but it takes an extra phase. The in-house encryptor replaces original filenames with random strings and the encrypted file retains no record of the old name. Reconstruction draws on backup file catalogues and directory listings, database-internal schema and metadata, embedded document properties such as author, title and creation time, content fingerprints matched to business identifiers, and cross-referencing with upstream and downstream systems. This workload is routinely underestimated, so we itemise it separately during assessment.
Our ESXi host still boots normally - does that mean the damage is limited?
No. The DragonForce ESXi encryptor carries whitelists of system directories and files specifically to keep the host bootable, encrypting only VM files under /vmfs/volumes so the victim can still log in, read the note and keep the negotiation channel open. A booting host says nothing about the damage - typically every VM on it is down. The priority is to stop all writes to the datastore immediately, avoid re-initialising it or creating new VMs, and establish the state of snapshots and backups first.
The attackers got in by impersonating IT over the phone - can that still be traced?
Yes, but the forensic focus differs from a malware-led intrusion. Social-engineered access often leaves no malicious binaries on endpoints; the evidence sits in service-desk tickets and call records, password and MFA reset audit logs, device registration and sign-in geography in the identity platform, VPN sessions and odd-hours logins, and the first appearance of suspicious administrative actions. We chain these with later lateral movement into a single timeline, which both locates the entry point and establishes the exfiltration scope for regulatory reporting.
Sources
- DragonForce Ransomware Group — Group-IB
- Ransomware Spotlight: DragonForce — Trend Micro
- Deep dive into DragonForce ransomware and its Scattered Spider connection — BleepingComputer
- Who are DragonForce and how the ransomware affiliate ecosystem is changing — Sophos
- Tracking a Dragon: investigating a DragonForce-affiliated ransomware attack — Darktrace
External links are provided for reference only. The content is published by third parties and does not represent our position.
Updated