A new Mini Shai-Hulud worm variant named Trinitite was detected targeting the npm package @7nohe/openapi-react-query-codegen. The attacker exploited a GitHub workflow vulnerability allowing any user to trigger npm publish via pull-request comments, resulting in rapid publication of malicious package versions. The worm uses obfuscation techniques to execute even when scripts are disabled, steals credentials from multiple developer and cloud platforms, and exfiltrates data via GitHub commits. It establishes persistence through systemd services and includes a destructive token revoke mechanism that can wipe user directories. This campaign is linked to the TeamPCP threat actor and appeared shortly after arrests of related suspects in Australia.
A Magecart campaign uses EtherHiding techniques to perform card-skimming on e-commerce platforms. Attackers compromise legitimate online storefronts, mainly WooCommerce, but also PrestaShop, Magento, and WordPress sites, injecting malicious loaders disguised as Google Tag Manager code. The skimming payloads are staged inside Ethereum blockchain smart-contract storage, making detection more difficult. Over 40 websites across 15 countries have been impacted since April 2026. The infrastructure involves 144 Sepolia contracts controlled by a single wallet, with 20 distinct contracts and skimmer-hosting domains identified. The malicious code appears as normal analytics but steals credit card data from shoppers.
Between January and April 2026, a coordinated social engineering operation named Spring Ring leveraged external Microsoft Teams accounts to impersonate IT help desk personnel, targeting over 150 employees across at least 10 companies. Adversaries initiated voice phishing calls to coerce victims into executing remote monitoring and management tools or custom malware. In advanced variants, attackers transitioned from vishing to NTLM relay attacks targeting domain controllers. Two distinct campaigns were observed: Campaign A utilized RMM tools and obfuscated PowerShell-based RATs, while Campaign B employed tailored cloud infrastructure with PetitPotam exploitation for domain-level compromise. The operation demonstrates the weaponization of trusted collaboration platforms as primary attack vectors, exploiting the trust gap in SaaS applications.
An attacker exploited a flawed GitHub Actions workflow in the npm package @7nohe/openapi-react-query-codegen to publish 10 malicious versions. The vulnerability stemmed from an issue_comment trigger lacking author-association checks, allowing arbitrary users to trigger npm publishing from forked repositories. The malicious payload used multiple execution methods, including a binding.gyp file exploiting node-gyp's Python evaluation and a preinstall hook, ultimately delivering a large obfuscated script that downloaded Bun v1.4.0. This attack aligns with the Mini Shai-Hulud toolkit previously seen in other supply chain compromises. All malicious versions have been removed from npm.
An investigation into malware delivery infrastructure reveals an operator using GitHub repositories to stage malicious loaders and RAT payloads. Starting from commit metadata, researchers traced an email address to a compromised machine via stealer log databases. The infected workstation exposed a complete operational pipeline including multiple RAT families (AsyncRAT, DcRat, Remcos, XWorm), phishing templates impersonating Colombian government institutions, bulk email software, and commercial crypter services. The operator maintains delivery infrastructure across GitHub, Bitbucket, AWS S3, and DuckDNS for command-and-control. Phishing campaigns target Colombian organizations using judicial notification and traffic violation lures with password-protected archives. The investigation demonstrates how infrastructure analysis and operational security failures can expose entire malware production workflows beyond individual samples.
This threat involves a sophisticated malware campaign using fake MP4 files as carriers for encrypted malicious payloads. The MP4 files appear legitimate to basic file-type checks but contain encrypted data in ISO Base Media File Format uuid extension boxes instead of playable video. The infection chain starts with PowerShell scripts delivered via Cloudflare-fronted infrastructure, which perform environment checks before retrieving the malicious carrier file. A secondary script extracts and decrypts an embedded PowerShell payload that silently installs the NetSupport Manager remote administration tool. The attack infrastructure includes multiple live endpoints across several autonomous systems, mainly in Frankfurt and Los Angeles, with rapid registration of command-and-control gateways. The toolkit uses Russian-language business site decoys and frequently rotates carriers without backward compatibility.
A Chinese-language executable disguised as a resume claiming to be from a Beijing Institute of Technology graduate student delivers SNOWLIGHT and VShell RAT. The infection chain uses a custom Go loader with sandbox detection, CPU checks, and sleep-timer evasion before downloading a legitimate DOCX decoy. It then executes a 1,454-byte SNOWLIGHT stager that connects to infrastructure, downloads 4.65 MB of data, XOR-decodes it with key 0x99, and launches a fileless VShell RAT. The lure targets mainland Chinese academics in electrical engineering, power systems, or AI research, providing operators with command execution, file access, screen capture, and lateral movement capabilities. The delivery uses panel-generated SNOWLIGHT components now widely available in cracked VShell releases, making actor-level attribution unreliable beyond identifying an unattributed operator using commodity tooling with China-oriented academic targeting.
In July 2026, multiple APT campaigns targeted South Korean entities using spear phishing emails with malicious LNK files. Seven distinct attack types employed various techniques such as PowerShell scripts, AutoIt programs, DLL side-loading, and curl.exe downloads. Malware was distributed via platforms like GitHub, Google Drive, and Dropbox, disguised as legitimate documents or resumes. The campaigns deployed backdoors, infostealers, keyloggers, and XenoRAT malware to exfiltrate sensitive information and maintain persistence through Task Scheduler entries. Communication with command and control servers used PubNub channels with Base64-encoded data. The attacks focused on deceiving victims with work-related content to execute malicious payloads.
A misconfigured open directory on IP address 86.53.111[.]212:8080 exposed critical details of an active cybercrime operation, including Moobot botnet source code, denial of service tools with attack records, and a fraudulent Chinese identity verification service. The exposed directory also contained StresD Pro+, a multi-user DDoS panel with 16 registered accounts and 32 recorded attacks on the collection date. The recovered Moobot source code revealed a previously unknown dormant download-and-execute functionality that represents the most plausible mechanism behind APT28's repurposing of Moobot for deploying malware to compromised devices. Despite a 2024 court-authorized disruption by the U.S. Department of Justice, Moobot remains active as of August 2026, with one active C2 server observed conducting over 500 short-duration attacks throughout the month, consistent with DDoS-as-a-service operations.
A campaign identified by the Socket Threat Research team involves 19 malicious browser extensions (18 for Chrome, 1 for Edge) active since February 2024. These extensions deliver a modular malware framework that communicates with command and control servers via WebSocket, strips Content Security Policy headers, and uses cross-site scripting (XSS) injection to execute payloads. The primary goal is to steal cryptocurrency wallet secrets and drain crypto assets, as well as harvest credentials. The threat actor either creates malicious extensions from scratch or compromises legitimate extensions with existing user bases, such as the 'Enable Right Click & Copy' extension with tens of thousands of users. The campaign, named 'Superior,' shows sophisticated and evolving capabilities targeting multiple cryptocurrency platforms and exchanges.
Three distinct implants—SPEAKINGSTONE, DARKLANTERN, and ENDLESSDOORS—have been discovered embedded in ZBT router firmware distributed through a global supply chain reaching the United States, Canada, Australia, Philippines, Germany, and Russia. DARKLANTERN is an unauthenticated UDP backdoor listening on port 9992, offering root shell access from the internet with trivial MAC address bypass. SPEAKINGSTONE is a phone-home surveillance implant that beacons to ZBT's cloud infrastructure, capable of DNS hijacking, ISP credential theft, and remote command execution. A sinkholed backup domain revealed 392 devices, 390 located in China, primarily on China Mobile's network. Internet scans identified 203 DARKLANTERN instances across 22 countries. These implants use plaintext protocols without authentication, making them hijackable by any network adversary. The affected hardware appears in white-labeled consumer products sold through mainstream retailers, representing embedded surveillance capabilities in commerciall...
In July 2026, Microsoft Threat Intelligence uncovered a supply chain attack targeting the official AsyncAPI NPM organization. Attackers published malicious versions of multiple packages under the trusted AsyncAPI namespace, exploiting developer dependencies to distribute malware. The compromised packages deployed a multi-stage Remote Access Trojan through obfuscated lifecycle hooks that executed during routine build workflows. Upon installation, the malware retrieved second-stage payloads from IPFS gateways, established persistence on infected systems, and initiated command-and-control communications with external infrastructure. The attack leveraged trusted build automation and dynamic package retrieval via npx to bypass traditional security controls, affecting developers executing version-pinned tasks in their CI/CD pipelines.
This analysis discusses how poorly designed AI guardrails in security operations can unintentionally aid attackers by impeding defensive actions. Overly restrictive or inflexible AI filters controlled by third-party providers may cause delays or refusals in security investigations, giving adversaries more time to complete their objectives. The author recommends that security teams maintain operational sovereignty over guardrails, allowing customization and temporary adjustments to safeguards to better align with specific threat models. This flexibility is essential to prevent attackers from exploiting rigid controls to disrupt incident response processes.
Between late September 2025 and early April 2026, the threat group BlueDelta conducted espionage campaigns targeting government and diplomatic organizations in Romania, Spain, and Türkiye. They deployed HOOKEDGE, a lightweight Windows batch-script backdoor, via macro-enabled Word documents with diplomatic-themed lures. HOOKEDGE shares code and tradecraft overlap with the HEADLACE backdoor and abuses legitimate webhook services for command-and-control, payload staging, and data exfiltration. The implant was continuously refined to evade sandbox detection and adapt to webhook service constraints. BlueDelta used a tiered operational model, deploying second-stage payloads with shorter beaconing intervals for high-value targets while preserving initial access infrastructure.
A China-linked cyber espionage infrastructure provider operates a multi-component 'quartermaster' system that offers reconnaissance, proxy orchestration, and traffic routing services to Chinese state-sponsored actors. The infrastructure includes QScan for reconnaissance, Fast Labyrinth for encrypted relay networks using commercial proxy services, QTRouter for proxy access management, and QTProxy for operational node control. It targets research universities, defense networks, government agencies, and critical infrastructure worldwide, with notable focus on the U.S., U.K., and Asia-Pacific regions. The operation leverages commercial proxy services designed to bypass China's Great Firewall, enabling multiple threat actors to maintain anonymity and coordinate operations via shared infrastructure. This represents an advanced evolution in state-enabled cyber espionage capabilities.
The Iranian-linked cyber-espionage group Tortoiseshell has expanded its malware toolkit with a new reverse SSH tunneling utility disguised as wtsapi32.dll and a C++ backdoor similar to TWOSTROKE malware. The SSH tunnel uses the Windows OpenSSH client to connect to command-and-control servers. The backdoor supports executing files, shell commands, in-memory DLL execution, and file manipulation. Infrastructure analysis shows domains linked to multiple countries including the UAE, Saudi Arabia, UK, Belgium, Canada, Australia, Japan, and the United States, indicating broader targeting beyond the Middle East and US defense and military sectors. The group has been active since 2018 and maintains persistent infrastructure despite domain suspensions. No known exploits in the wild or patches are reported.
This is a sophisticated global phishing campaign that abuses legitimate Remote Management and Monitoring (RMM) software to gain unauthorized remote access. Initially targeting Canadians with fake Canada Revenue Agency tax documents, it has expanded to 46 countries, with nearly half of the activity in the United States. Attackers impersonate trusted organizations and use advanced techniques such as password-protected archives, browser fingerprinting, and Telegram-based victim filtering. The campaign uses rapidly rotating infrastructure hosted mainly on Vercel, leveraging legitimate TLS certificates and domain reputations. Signed commercial RMM tools are abused, making traditional signature-based detection ineffective. The campaign has been active steadily since January 2026.
ERMAC and HookBot are two branches of one Android banking trojan sold as a service, forking from shared code originating with Cerberus. A copy of the builder, Laravel backend, and React panel leaked in August 2025, enabling unrelated operators to deploy panels with default credentials and keys still in place. The lineage runs Cerberus to ERMAC to Hook, confirmed through source code analysis showing identical database migrations and network protocol structures. HookBot added VNC remote control and 38 new commands while maintaining ERMAC's core. The leaked source includes a Docker stack, Obfuscapk builder, and IP-whitelist firewall that hides panels but leaves the builder port exposed. Operators target 484 apps across 40+ countries including Japanese banks, Brazilian financial institutions, Turkish banks, and cryptocurrency wallets. Detection artifacts survive in builder obfuscator flags and favicons, while panel titles remain easily changed.
Analysis of over 400 AI-enabled malware samples shows that most remain confined to research and sandbox environments, with only a small fraction observed on protected endpoints across three countries. These samples span five malware families including FunkSec ransomware and Oyster backdoor. Existing behavioral detection, cloud sandboxing, and endpoint analytics successfully detect and block all observed samples. The AI component primarily accelerates malware development rather than enabling evasion of defenses. Distribution patterns are opportunistic rather than targeted.
A sophisticated multi-stage loader dubbed PavinLoader has been identified across multiple distribution campaigns including ClickFix attacks, fake software downloads, and malicious RenPy games. The loader employs heavily obfuscated .NET DLLs, abuses legitimate Windows tools like MSBuild, and utilizes EtherHiding technique to retrieve command-and-control domains via blockchain. The infection chain consists of four main stages: a Loader DLL performing anti-forensics, an EtherHiding Loader obtaining C2 infrastructure, an Anti-Analysis DLL checking for virtualized environments, and a PE Loader delivering final payloads including Amatera Stealer. Evidence suggests PavinLoader may be offered as a Loader-as-a-Service, with common artifacts found across over 200 related files. The campaigns demonstrate sophisticated evasion techniques including custom obfuscation, API hashing, and extensive anti-analysis checks targeting virtualized environments and specific geographic regions.
