Threats Feed

The RedKitten campaign, observed in early 2026, targets Iranian interests, specifically NGOs and individuals documenting human rights abuses during the "Dey 1404" protests. Assessing the actor as Iranian state-aligned, researchers identified "SloppyMIO," a modular .NET implant likely developed with AI assistance. The attack chain utilizes spearphishing with "shock lures" regarding execution lists to deliver malware via AppDomainManager injection. The threat actor leverages legitimate infrastructure, using GitHub as a Dead Drop Resolver for steganographic configuration, Google Drive for payload hosting, and Telegram for command and control. This campaign highlights the growing use of LLMs in rapid malware development and the exploitation of civil unrest for targeted surveillance in Iran.

APT42 utilizes TAMECAT, a modular PowerShell-based backdoor, to target high-value senior defense and government officials. Israel’s National Digital Agency reports that the group employs social engineering to gain initial access. The infection chain begins with a VBScript that profiles antivirus software via WMI to determine whether to deploy PowerShell or Command Shell downloaders. TAMECAT features sophisticated capabilities, including screen capture, Chrome data collection, and Microsoft Edge remote debugging. It leverages legitimate services like Telegram and Discord for Command and Control (C2). Data is encrypted via AES and exfiltrated to domains such as glitch[.]me, demonstrating APT42's focus on stealth and persistent espionage operations.

The report analyzes the evolution of RustyStealer (also referenced as RustyWater or Archer RAT), a Rust-based post-compromise implant observed in MuddyWater-attributed activity. By correlating build artefacts, compiler metadata, dependency drift, TLSH similarity, and API-level changes, the analysis reconstructs a development timeline from early baseline builds to a more mature v2.0 architecture. Rather than linear feature growth, the samples reveal experimentation, rollback, and consolidation. A short-lived asynchronous I/O refactor was abandoned in favor of improved stability, while later versions emphasize stealth through native NT API usage, runtime string obfuscation, and restored host fingerprinting.

The report analyzes a newly observed MuddyWater malware sample that exposes extensive build and development artifacts due to improper binary stripping. Delivered via a malicious Word document containing VBA macros, the payload reconstructs and executes a Rust-based executable on disk. Analysis of leftover strings reveals detailed insights into the actor’s development environment, including a Windows-based build host, MSVC Rust toolchain, local Cargo usage, and a recurring username embedded in build paths. These artifacts indicate locally compiled tooling with minimal release hardening and weak OPSEC. The findings highlight how developer mistakes can provide durable fingerprints for clustering, campaign tracking, and long-term threat hunting, beyond traditional infrastructure indicators.