9c428a30-4

ConfirmedSecurity569 words3 min decode

Suspected North Korean Hackers Hid Malware Command Servers in Solana Memos

Cisco Talos tied group UAT-12643 to malware that stored C2 addresses in Solana memos, survived full Windows resets, and targeted crypto holders, gamers and VoIP users.

Suspected North Korea malware uses Solana memos to switch servers
WitnessSuspected North Korea malware uses Solana memos to switch serversTho-Ge / pixabay

Outputs

  1. Cisco Talos published the research on November 18, tracking the group as UAT-12643.

  2. Overlord RAT used encrypted Solana transaction memos to update its C2 server addresses.

  3. Talos identified 35 commands, including keylogging, webcam capture and lateral movement.

  4. North Korea attribution carries low confidence; the tool and technique have other users.

  5. The malware survived a full Windows system reset via the Windows Recovery Environment.

Cisco Talos on November 18 published research on a financially motivated hacking group, tracked as UAT-12643, that stored encrypted command-and-control server addresses inside Solana transaction memos. The technique let operators rotate their infrastructure without redeploying malware on infected machines. Talos attributes the group to a North Korea-aligned advanced persistent threat with low confidence — a caveat the report itself stresses.

The toolset centers on Overlord, a Go-based remote access trojan. Infected machines queried an attacker-controlled Solana wallet, where operators had placed encrypted C2 addresses in transaction memos. When a server address changed, the malware simply read the new one from the chain.

How does the blockchain change the attack?

Blockchain-based C2 removes a classic weakness of malware: hardcoded server addresses that defenders can block and researchers can sinkhole. By publishing addresses on Solana, the operators gained a tamper-resistant, always-available channel that survives takedowns of any single host. Talos also cited prior use of a similar Solana-based C2 technique by North Korea's Lazarus Group, though the researchers noted other actors have used both Overlord and the blockchain method.

The report identified 35 commands within Overlord. Capabilities include:

  • keystroke logging;
  • screen, webcam and audio capture;
  • file and process control;
  • script execution;
  • lateral movement across internal networks;
  • in-memory plugin delivery.

Removal resistance was a design priority, not an afterthought. Terminating the malware triggered a Windows Blue Screen of Death, and its foothold in the Windows Recovery Environment — the component used to repair or reset a PC — allowed it to persist. Talos said this ensured "the implant survives a full system reset." In plain terms: reinstalling the operating system was not enough to evict the attacker.

Who was in the crosshairs?

Lures and payloads suggest three target groups: cryptocurrency holders, gamers, and users of internet telephony software. Decoy files included a fabricated financial statement, a tampered Viber installer, and documents named to resemble an internet calling application.

The attackers supplemented Overlord with Good Game, a freely available Python-based infostealer that Talos confirmed the group did not build itself. That stealer harvests browser data, Discord tokens, cryptocurrency wallets and gaming details. Its file-scanning routines searched in English, French, Spanish and Portuguese, which Talos read as a sign that intended victims may sit in Europe.

A separate clipboard hijacker completed the monetization chain. It swapped copied cryptocurrency wallet addresses with attacker-controlled addresses during copy-and-paste operations — the moment a user pastes what they believe is a recipient address, funds route to the thief.

What does the attribution actually rest on?

Very little that is direct. Talos's North Korea assessment draws on two threads: prior deployment of Overlord by UNK_DeadDrop, a cluster Talos previously judged probably North Korean, and the Lazarus-linked Solana C2 precedent. The researchers themselves flag that other actors have used both the tool and the technique, which narrows the evidentiary value of either signal.

One operational security lapse gave Talos its clearest visibility: the actor left its file server publicly reachable, exposing parts of the toolset and live C2 settings. Attribution, in other words, rests partly on the attackers' own mistake rather than on network telemetry alone.

For exchanges, custodians and security teams, the operational lesson is narrow but real. Address poisoning via clipboard hijacking and blockchain-hosted C2 both evade traditional indicators, and monitoring for anomalous Solana memo activity now belongs on threat-hunting checklists.

via en.cryptonomist.ch (Source)

More from Elena Vasquez

Elena Vasquez

Show full bio

Staff writer covering regulation, policy and their impact on crypto and web3 industry at Mempool Brief.

1,647 articles