Threats Tagged 'post-exploitation'

PEEP is a Chromium-based post-exploitation toolkit disguised as 'Smart Bookmarks' that transforms Chrome and Edge browsers into persistent backdoors. Requiring prior administrative access, it bypasses Web Store security by forging Chromium integrity values and uses native-messaging to extend beyond browser telemetry to full host-level command execution. The malware beacons every 30 seconds over unencrypted HTTP, exfiltrating browsing history, cookies, credentials, and session data. Built on the open-source RedExt framework with significant enhancements, PEEP combines browser surveillance with filesystem manipulation, shell command execution, and process discovery. The infrastructure exposed development artifacts including source code, signing keys, and Traditional Chinese testing logs, indicating a Chinese-speaking operator using AI-assisted development. The toolkit demonstrates sophisticated persistence through HMAC forgery, enterprise policies, and ScriptCache manipulation.

In this article AI workloads are becoming high-value control points Case study 1: LiteLLM gateway compromise Case study 2: RAGFlow compromise Case study 3: Kestra compromise Mitigation and protection guidance MITRE ATT&CK techniques observed References Learn more AI is creating a new layer of enterprise infrastructure. Gateways, retrieval platforms, orchestration services, and containerized runtimes now sit between users, applications, data, and models. These systems concentrate credentials, data access, model connectivity, and execution privileges, making them some of the most powerful components in the AI stack. That concentration of trust is also creating new opportunities for attackers. In recent investigations, Microsoft observed activity targeting three distinct AI workloads: a LiteLLM gateway, a RAGFlow deployment, and a Kestra workflow environment. The intrusion paths varied, but the objectives were strikingly similar. Attackers sought to steal credentials, establish persistence, and monetize compromised compute resources. The individual techniques matter, but the broader pattern matters more. Across these cases, attackers treated AI infrastructure as a control plane where credential theft, host compromise, and downstream data access can converge. As organizations continue to deploy AI systems, these platforms are becoming high value targets that deserve the same security scrutiny as other critical enterprise infrastructure. AI workloads are becoming high-value control points The campaign-level signal extends beyond one product. The targeted workloads served different functions, but each exposed assets that could support follow-on abuse, including model-provider keys, proxy-issued virtual keys, database connection strings, tenant configuration, workflow execution, or host compute. Post-compromise behavior varied by workload role. Defenders should inventory exposed AI management surfaces, restrict administrative access, and monitor for gateway-originated execution and secret access. Three observed compromises across AI workloads AI workload Observed activity Attacker objective LiteLLM Observed attacker activity : Python droppers, runtime secret harvesting, PostgreSQL collection, miner deployment, and persistence activity from the LiteLLM gateway context. Microsoft assessment: Initial access likely occurred through exploitation of the exposed LiteLLM gateway surface, consistent with the vulnerability chain involving CVE-2026-42271 and CVE-2026-48710. Credential theft, backend database access, durable host access, and compute monetization. RAGFlow Observed attacker activity : Possible SSRF-style reconnaissance followed several days later by code execution, application-path modification, and placement of a Python hook in the TenantLLM credential-configuration flow. Public research: Describes multiple RAGFlow execution paths; Microsoft does not attribute this intrusion to a specific vulnerability. Intercept newly configured LLM provider credentials and model metadata. Kestra Observed attacker activity : Workflow-origin shell execution, Docker and container-environment discovery, XMRig deployment, and follow-on data collection. Microsoft assessment: Initial access likely involved exploitation of the exposed Kestra orchestration surface, with CVE-2026-49869 providing relevant public vulnerability context. Secret discovery, container-level access, data collection, and rapid compute monetization. Case study 1: LiteLLM gateway compromise Framework role and affected runtime context LiteLLM is commonly deployed as a proxy or gateway between applications and model providers. In that position, the service may hold or retrieve model-provider keys, LiteLLM master keys, virtual-key records, database connection strings, routing configuration, and tenant policy data. Command execution in the gateway runtime therefore exposed a process context close to AI routing and credential material. Figure 1. LiteLLM gateway compromise – attack chain…

A new post-exploitation red team tool named Splinter has been discovered on customer systems through Advanced WildFire's memory scanning capabilities. Developed in Rust programming language, Splinter is exceptionally large at around 7MB due to statically linked libraries. The tool uses a JSON configuration structure containing implant ID, C2 server details, and operational parameters. It operates through a task-based model with capabilities including Windows command execution, remote process injection, file upload/download, cloud service information gathering, and self-deletion. Communication with the C2 server occurs via HTTPS using specific URL paths for task synchronization, heartbeat connections, and file transfers. While not as sophisticated as Cobalt Strike, Splinter represents a growing variety of penetration testing tools that could potentially be misused by threat actors.

A sophisticated ClickFix campaign was observed in April 2026 deploying PySoxy, a decade-old open-source Python SOCKS5 proxy tool, to establish encrypted proxy access on compromised hosts. The attack chain begins with social engineering that tricks users into executing obfuscated PowerShell commands, which then establishes scheduled task persistence and deploys an in-memory PowerShell-based command-and-control agent. Following domain reconnaissance activities, attackers deploy PySoxy to create a redundant encrypted access channel. The persistence mechanism continues attempting re-execution even after initial connections are blocked, demonstrating how single ClickFix executions can evolve into modular post-exploitation chains. This development represents a significant evolution from simple one-time execution to durable access with multiple redundant pathways, requiring comprehensive remediation beyond blocking initial callbacks.

MuddyWater APT has launched Operation Olalampo, targeting organizations in the MENA region. The campaign involves new malware variants, including a Rust backdoor called CHAR, downloaders GhostFetch and HTTP_VIP, and an advanced backdoor GhostBackDoor. Notably, the group is using Telegram bots for command-and-control, revealing insights into their post-exploitation tactics. The operation, first observed on January 26, 2026, shows tactical and technical overlaps with previous MuddyWater activities. Key discoveries include potential AI-assisted malware development and infrastructure reuse dating back to October 2025. The campaign aligns with ongoing geopolitical tensions and provides valuable information on the threat actor's evolving techniques.

MediumMalwareUnited Arab EmiratesSaudi ArabiaQatar+7#ghostbackdoor#ai-assisted#operation olalampo

An exposed open directory on a command and control server revealed a complete deployment of the BYOB (Build Your Own Botnet) framework. The multi-stage infection chain targets Windows, Linux, and macOS platforms, implementing seven persistence mechanisms. The malware includes extensive post-exploitation capabilities such as keylogging, packet capture, and email harvesting. Analysis uncovered a modular design with encrypted C2 communications and infrastructure reuse across multiple regions. Two nodes also hosted XMRig cryptocurrency miners, indicating additional monetization efforts. The campaign has been operational for approximately 10 months, demonstrating geographic and provider diversification in its infrastructure.

MediumMalwareGermanyFranceUnited Kingdom+4#persistence#cryptomining#byob

A malicious npm package named 'https-proxy-utils' was discovered containing a post-install script that downloads and executes the AdaptixC2 post-exploitation framework agent. The package impersonated legitimate utilities and cloned functionality to evade detection. It targets Windows, Linux, and macOS platforms with tailored payload delivery methods. Once installed, the AdaptixC2 agent enables remote access, command execution, and persistence, posing significant risks to affected systems. This incident exemplifies the increasing abuse of open-source software supply chains as attack vectors. No known exploits in the wild have been reported yet, but the threat remains active and multi-platform. The malicious infrastructure is hosted on the domain cloudcenter.top, which serves payloads for different OS architectures. European organizations relying on npm packages for development or deployment are at risk, especially those with mixed OS environments. Mitigation requires vigilant package vetting, monitoring for suspicious post-install scripts, and network controls to block malicious domains.

MediumMalwareGermanyFranceNetherlands+5#adaptixc2#post-exploitation#npm

AdaptixC2, an open-source post-exploitation and adversarial emulation framework, has been observed being used in real-world attacks. This versatile tool allows threat actors to execute commands, transfer files, and perform data exfiltration on compromised systems. Its open-source nature enables easy customization, making it highly flexible and dangerous. The framework supports sophisticated tunneling capabilities, modular design with extenders, and various beacon agent formats. Two infection scenarios were analyzed: one using social engineering via Microsoft Teams, and another likely involving AI-generated scripts. The increasing prevalence of AdaptixC2 in attacks, including its use alongside ransomware, highlights the growing trend of attackers leveraging customizable frameworks to evade detection.

MediumCampaignGermanyUnited KingdomFrance+7#data exfiltration#c2 framework#open-source

A threat actor is using compromised WordPress websites to distribute a malicious version of NetSupport Manager Remote Access Tool (RAT). The attack chain involves phishing campaigns, website compromise, DOM manipulation, and a fake CAPTCHA page. The malware is delivered through a batch file that downloads and executes NetSupport Client files. Post-infection, the attacker uses NetSupport's features for reconnaissance and further exploitation. The attack utilizes various JavaScript files and DOM manipulation techniques to evade detection. Multiple IP addresses and domains associated with the attack infrastructure have been identified, primarily linked to hosting providers in Moldova.

MediumMalwareGermanyFranceUnited Kingdom+4#post-exploitation#netsupport rat#fake captcha

An initial access broker exploited leaked Machine Keys on ASP.NET sites to gain unauthorized access to organizations. The group, tracked as TGR-CRI-0045, targeted industries in Europe and the U.S.including finance, manufacturing, and technology. They used ASP.NET View State deserialization to execute malicious payloads in server memory, minimizing forensic artifacts. The attackers deployed post-exploitation tools for persistence and privilege escalation. The campaign began in October 2024, with increased activity from January to March 2025. Organizations are advised to review and remediate compromised Machine Keys following Microsoft's guidance. The threat group is possibly linked to Gold Melody based on overlapping indicators and tactics.

MediumMalwareGermanyFranceUnited Kingdom+7#in-memory execution#updf#privilege escalation