Cyware Weekly Threat Intelligence, May 12–16, 2025

The Good

Smart heat pumps are getting a cybersecurity upgrade. Under the U.K.'s new Smart Secure Electricity Systems Programme, devices up to 45 kW must now comply with ETSI EN 303 645 standards. The move aims to protect both consumers and the national grid from potential cyber threats. Blockchain just got a rulebook refresh. The BSSC has rolled out four new standards. Together, they offer a structured approach to tackling security across blockchain infrastructures.

• Smart heat pumps in the U.K are now subject to new cybersecurity rules under the Smart Secure Electricity Systems Programme. These rules require compliance with the ETSI EN 303 645 standard, ensuring better protection for consumers, their data, and the national grid. The regulations apply to heat pumps with a capacity of up to 45 kW and aim to address risks such as cyberattacks and grid instability. Manufacturers must prepare for these changes by adopting secure-by-design practices and ensuring their devices meet the required standards before enforcement begins in late 2026 or early 2027.
• The EU has launched She@Cyber, a free beginner cybersecurity training program targeting women and underrepresented groups to address the cyber skills gap. The program, supported by Erasmus+ and coordinated by Vernian RTI, offers industry-recognized credentials based on ISACA’s Cybersecurity Fundamentals Certificate. SMEs, startups, and microenterprises are encouraged to use the program to find cybersecurity talent.
• The Blockchain Security Standards Council (BSSC) has launched four new security standards aimed at enhancing blockchain security. These include the Node Operation Standard (NOS) for secure node operations, the Token Integration Standard (TIS) for safe digital asset integration, the Key Management Standard (KMS) for secure cryptographic key handling, and the General Security and Privacy Standard (GSP) for baseline security and privacy requirements. 
• German police successfully dismantled the cryptocurrency swapping service eXch, in a coordinated operation with the Frankfurt am Main Public Prosecutor's Office. The Federal Criminal Police Office (BKA) seized €34 million ($38 million) in various cryptocurrencies, including Bitcoin, Ether, Litecoin, and Dash, marking one of the largest seizures in their history. The operation targeted eXch due to its involvement in extensive money laundering activities, with an estimated $1.9 billion laundered since its inception in 2014. 

The Bad

APT37 is back with another Dropbox-powered espionage play. In Operation: ToyBox Story, the North Korean group used fake national security event invites to deliver RoKRAT malware via ZIP archives. Two ransomware gangs, BianLian and RansomExx, are now exploiting the same SAP NetWeaver vulnerability as Chinese APTs. Using CVE-2025-31324, the attackers deliver PipeMagic alongside privilege escalation flaws like CVE-2025-29824. Swan Vector is the latest addition to a string of APT campaigns zeroing in on East Asia’s research and engineering sectors. 

• An investigation revealed the presence of SnipVex and XRedRAT malware in Procolored's software downloads, including Win32.Backdoor.XRedRAT.A and MSIL.Trojan-Stealer.CoinStealer.H.Malware. The XRed backdoor, identified in the PrintExp.exe sample, has existed since at least 2019 and allows malicious activities like keylogging, file downloads, and executing commands. The SnipVex virus, a clipbanker, diverts cryptocurrency transactions by replacing clipboard BTC addresses and infects executable files. The malware’s impact included 39 infected files and a BTC address linked to the attacker, which received approximately 9.3 BTC (~$100,000).
• North Korean APT37 launched Operation: ToyBox Story, using spear phishing disguised as national security forum invitations. Dropbox was used as a delivery and C2 channel, following a pattern of abusing trusted cloud platforms. The operation involved two cases, both delivering RoKRAT malware through malicious .lnk files in ZIP archives. RoKRAT is a remote access trojan that collects system information and communicates with C2 servers over Dropbox. The data exfiltration process involves multi-layer encryption before transmission.
• BianLian and RansomExx have exploited a security flaw (CVE-2025-31324) in SAP NetWeaver, leveraging it to deploy malicious payloads like the PipeMagic trojan. Evidence links BianLian to incidents through IP addresses and infrastructure associated with their previous activities. PipeMagic was used alongside a Windows CLFS privilege escalation vulnerability (CVE-2025-29824) in targeted attacks across multiple countries. The trojan was delivered via web shells after exploiting the SAP NetWeaver flaw, with subsequent attacks involving Brute Ratel C2 framework and exploitation of CLFS vulnerability through inline assembly.
• ESET researchers have uncovered Operation RoundPress, a cyberespionage campaign attributed to the Sednit group (aka Fancy Bear, APT28), targeting webmail servers via XSS vulnerabilities. The campaign expanded from targeting Roundcube in 2023 to include Horde, MDaemon, and Zimbra in 2024. Sednit used a zero-day XSS vulnerability (CVE-2024-11182) in MDaemon, which was patched in version 24.5.1. The primary targets are governmental entities and defense companies in Eastern Europe, with additional victims in Africa, Europe, and South America. The attack chain begins with spearphishing emails that exploit XSS vulnerabilities to execute malicious JavaScript payloads. SpyPress payloads (e.g., SpyPress.HORDE, SpyPress.MDAEMON, SpyPress.ROUNDCUBE, and SpyPress.ZIMBRA) are used for credential theft, email exfiltration, and bypassing security measures like 2FA. 
• Cybersecurity researchers have uncovered a phishing campaign distributing Horabot malware, targeting Windows users in six Latin American countries: Mexico, Guatemala, Colombia, Peru, Chile, and Argentina. It spreads through emails disguised as invoices, using malicious HTML files to download payloads. The malware employs VBScript, AutoIt, and PowerShell for credential theft and lateral propagation via Outlook. It checks for antivirus software and virtual machines to evade detection while collecting sensitive information like IP addresses and usernames. Horabot also steals browser data and automates phishing emails, effectively creating a network of infected users.
• Seqrite Labs discovered the Swan Vector APT campaign targeting Taiwan and Japan, primarily focusing on educational and mechanical engineering sectors. The campaign uses a four-stage malware deployment process, starting with a malicious LNK file and progressing through various DLL implants to deliver Cobalt Strike shellcode. The attack chain begins with a decoy ZIP file containing a malicious LNK and a PNG-masqueraded DLL, executed via rundll32.exe. The Pterois implant performs API hashing and downloads further malware stages from Google Drive using OAuth credentials. The Isurus implant uses DLL sideloading to execute shellcode extracted from an encrypted file, employing techniques like API hashing and direct syscall execution. Cobalt Strike shellcode is decrypted and utilized for process injection, with infrastructure details indicating the use of specific IPs and URLs.
• Earth Ammit, linked to Chinese-speaking APT groups, conducted two campaigns—VENOM (2023-2024) and TIDRONE (2024)—targeting the drone supply chain in Taiwan. The VENOM campaign focused on software service providers using open-source tools, while TIDRONE employed custom-built malware like CXCLNT and CLNTEND. Both campaigns impacted various sectors, including military, satellite, and technology, primarily in Taiwan and South Korea. Both campaigns aimed to compromise trusted networks through supply chain attacks, leveraging vulnerabilities in upstream vendors to infiltrate downstream targets.
• Turkish espionage group Marbled Dust exploited a zero-day vulnerability in the Output Messenger app to spy on the Kurdish military in Iraq. The flaw, CVE-2025-27920, was a directory traversal vulnerability that allowed unauthorized access to sensitive data. Despite a patch released by the app's developer, Srimax, not all users updated their software, allowing the exploit to continue. Marbled Dust, aligned with Turkish interests, typically targets entities opposing the Turkish government. The group used DNS hijacking and typo-squatted domains to intercept credentials, deploying malicious files to maintain access and exfiltrate data.

New Threats

OtterCookie is getting smarter with every version. The North Korea-linked group WaterPlum is actively using the malware to breach financial institutions and crypto platforms worldwide. A PowerShell script posing as a helpful utility is the entry point for Chihuahua Stealer. This .NET-based info-stealer uses scheduled tasks for persistence and grabs browser credentials and crypto wallet data. Not every AI tool on Facebook is what it claims to be. Threat actors are luring users with fake content-generation apps, pushing Noodlophile malware to tens of thousands of victims.

• A new malware campaign utilizes a PowerShell-based loader to deploy the Remcos RAT, employing malicious LNK files and mshta.exe for execution. Delivered via malicious LNK files in ZIP archives, the attack utilizes mshta.exe to run obfuscated VBScript, bypassing Windows Defender. The malware modifies the Windows Registry for persistence, downloads multiple payloads, and executes code directly in memory using Win32 APIs. Remcos features modules for keylogging, webcam access, and credential theft from browsers, employing anti-analysis techniques to evade detection. The latest version includes enhanced functionalities for managing victim machines.
• WaterPlum, a North Korea-linked attack group, has been using the OtterCookie malware to target financial institutions and cryptocurrency operators globally. Since its introduction in September 2024, OtterCookie has evolved through multiple versions, with v3 featuring file upload capabilities and Windows support, while v4 adds stealer modules for credentials from Google Chrome, MetaMask, and Brave browser. Differences in coding suggest varying developers for these modules. The group’s activities have been notably observed in Japan, with ongoing updates to the malware.
• ETH Zurich researchers have discovered a new security vulnerability in Intel processors, named Branch Privilege Injection. This flaw exploits speculative technologies in CPUs, enabling attackers to bypass security barriers and access sensitive data from other users on shared processors. All Intel processors manufactured since 2018 are affected, including those in PCs, laptops, and data center servers. The vulnerability arises from brief race conditions during prediction calculations, allowing hackers to gradually read the entire memory contents byte by byte.
• Researchers discovered a malicious npm package, os-info-checker-es6, which uses Unicode steganography to hide its code and Google Calendar as a dynamic C2 dropper. The package was initially posed as benign, but later versions included obfuscated code to extract a payload and contact a remote server. It has been downloaded 2,001 times. Google Calendar is leveraged as a trusted intermediary to obscure attacker infrastructure, complicating detection and blocking efforts.Three other npm packages appear to be part of the same campaign, indicating a broader attack strategy.
• TransferLoader is a newly identified malware loader active since at least February 2025. It comprises three main components—a downloader, a backdoor loader, and a backdoor—each employing advanced anti-analysis and obfuscation techniques. TransferLoader has been observed delivering Morpheus ransomware and is capable of executing arbitrary commands, maintaining persistence, and updating its C2 infrastructure via IPFS.
• Chihuahua Stealer is a .NET-based infostealer identified through a deceptive PowerShell script shared via Google Drive. The malware employs a multi-stage payload chain, achieving persistence through scheduled tasks and targeting browser data and crypto wallet extensions. Stolen data is compressed into a ".chihuahua" archive and encrypted using AES-GCM, then exfiltrated over HTTPS while erasing local traces. Its techniques include Base64 encoding, hex-string obfuscation, and dynamic payload retrieval from fallback domains, demonstrating a sophisticated approach to evade detection.
• Threat actors are leveraging fake AI tools on Facebook to distribute the Noodlophile malware, targeting over 62,000 users. These tools masquerade as legitimate AI-powered content creation services, tricking users into downloading malicious files. The malware harvests sensitive data, including browser credentials and cryptocurrency information, through a complex infection chain initiated by a deceptive executable.