All news with #data exfil via tools tag

Chinese Hackers Turn ArcGIS Server into Year-Long Backdoor

🛡️ReliaQuest attributes a campaign to China-linked group Flax Typhoon that compromised a public-facing ArcGIS server by converting a Java Server Object Extension (SOE) into a gated web shell, maintaining access for over a year. The attackers embedded a hard-coded key and hid the backdoor in system backups to survive full system recovery. They uploaded a renamed SoftEther executable (bridge.exe), created a "SysBridge" service to persist, and used an outbound HTTPS VPN bridge to extend the victim network for covert lateral movement. Investigators observed credential theft, admin account resets, and extensive living-off-the-land activity to evade detection.

Malicious npm, PyPI and RubyGems Packages Use Discord C2

⚠️ Researchers at a software supply chain security firm found multiple malicious packages across npm, PyPI, and RubyGems that use Discord webhooks as a command-and-control channel to exfiltrate developer secrets. Examples include npm packages that siphon config files and a Ruby gem that sends host files like /etc/passwd to a hard-coded webhook. The investigators warn that webhook-based C2 is cheap, fast, and blends into normal traffic, enabling early-stage compromise via install-time hooks and build scripts. The disclosure also links a large North Korean campaign that published hundreds of malicious packages to deliver stealers and backdoors.

Stealit Infostealer Campaign Deploys via Fake VPN Apps

🛡️ FortiGuard Labs has identified a campaign distributing the Stealit infostealer via disguised game and VPN installers shared on file‑hosting sites and platforms like Discord. Attackers use Node.js Single Executable Apps (SEA) and PyInstaller bundles, heavy obfuscation and multiple anti‑analysis techniques to avoid detection. Once executed, Stealit harvests data from browsers, game clients, messaging apps and cryptocurrency wallets, and its operators rotate C2 domains while marketing the toolkit commercially.

Velociraptor Abuse Enables Stealthy Ransomware Campaigns

🔒 Researchers report that the open-source DFIR tool Velociraptor was abused by threat actors to maintain stealthy persistent access while deploying multiple ransomware families, including Warlock, LockBit and Babuk. Cisco Talos observed the activity in August 2025 and attributed the multi-vector operation to a China-linked cluster tracked as Storm-2603. Attackers exploited a vulnerable agent (v0.73.4.0) via CVE-2025-6264 to escalate privileges and persist; defenders are urged to verify deployments and update to v0.73.5 or later.

Indirect Prompt Injection Poisons Agents' Long-Term Memory

⚠️This Unit 42 proof-of-concept shows how an attacker can use indirect prompt injection to silently poison an AI agent’s long-term memory, demonstrated against a travel assistant built on Amazon Bedrock. The attack manipulates the agent’s session summarization process so malicious instructions become stored memory and persist across sessions. When the compromised memory is later injected into orchestration prompts, the agent can be coerced into unauthorized actions such as stealthy exfiltration. Unit 42 outlines layered mitigations including pre-processing prompts, Bedrock Guardrails, content filtering, URL allowlisting, and logging to reduce risk.

ClayRat Android spyware mimics popular apps to spread

📱 A new Android spyware campaign called ClayRat is tricking users by posing as well-known apps and services such as WhatsApp, Google Photos, TikTok, and YouTube and distributing APKs via Telegram channels and fraudulent websites. Researchers at Zimperium say they documented over 600 samples and 50 distinct droppers in three months, noting that some use a session-based installation and encrypted payloads to bypass Android defenses. Once installed, ClayRat can assume the default SMS handler, exfiltrate SMS and call logs, capture notifications and front-camera photos, make calls, send mass SMS for propagation, and communicate with C2 servers (recent versions use AES-GCM); Play Protect now blocks known variants.

Threat actors abusing Velociraptor in ransomware attacks

⚠️Researchers have observed threat actors leveraging the open-source DFIR tool Velociraptor to maintain persistent remote access and deploy ransomware families including LockBit and Babuk. Cisco Talos links the campaigns to a China-based group tracked as Storm-2603 and notes use of an outdated Velociraptor build vulnerable to CVE-2025-6264. Attackers synchronized local admin accounts to Entra ID, accessed vSphere consoles, disabled Defender via AD GPOs, and used fileless PowerShell encryptors with per-run AES keys and staged exfiltration prior to encryption.

Closing the Cloud Security Gap: Key Findings 2025 Report

🔒 The 2025 Unit 42 Global Incident Response Report shows that nearly a third of incidents investigated in 2024 were cloud-related, with 21% of cases directly impacting cloud assets. The article stresses the importance of the shared responsibility model and full, dynamic visibility to manage resource sprawl, misconfigurations and complex cloud-native architectures. It highlights identity misuse and overpermissioned accounts as frequent attack vectors and urges least privilege, credential rotation and robust logging. Palo Alto Networks recommends unified posture and response through Cortex Cloud and integration with Cortex XSIAM to reduce noise and automate remediation.

From Infostealer to PureRAT: Dissecting an Escalating Attack

🔍 Huntress Labs analyzed a multi-stage intrusion that began with a phishing ZIP and DLL sideloading and escalated to deployment of the commercial PureRAT backdoor. The operator combined bespoke Python loaders and a Python-based infostealer with compiled .NET loaders, process hollowing, AMSI/ETW tampering, and reflective DLL injection to evade detection. Final-stage configuration revealed a Vietnam-hosted C2 (157.66.26.209) and Telegram infrastructure linked to PXA Stealer, underscoring a shift from custom theft to a professional RAT.

Researchers Identify Architectural Flaws in AI Browsers

🔒 A new SquareX Labs report warns that integrating AI assistants into browsers—exemplified by Perplexity’s Comet—introduces architectural security gaps that can enable phishing, prompt injection, malicious downloads and misuse of trusted apps. The researchers flag risks from autonomous agent behavior and limited visibility in SASE and EDR tools. They recommend agentic identity, in-browser DLP, client-side file scanning and extension risk assessments, and urge collaboration among browser vendors, enterprises and security vendors to build protections into these platforms.

ClayRat Android Spyware Turns Phones Into SMS Hubs

🔔 A fast-evolving Android spyware campaign dubbed ClayRat has produced over 600 samples and 50 droppers in three months, researchers say. The malware is distributed via phishing sites and Telegram channels that impersonate popular apps like TikTok, YouTube and Google Photos to trick users into sideloading infected APKs. Once granted SMS privileges, ClayRat can read and send messages, harvest contacts and call logs, take front-camera photos, exfiltrate data to C2 servers, and automatically text malicious links to all contacts, turning each compromised device into a propagation hub.

ClayRat Android Spyware Campaign Targets Russian Users

🛡️Researchers at Zimperium zLabs have identified a rapidly evolving Android spyware campaign, dubbed ClayRat, targeting users in Russia via Telegram channels and phishing sites. The malware is distributed inside fake apps impersonating services such as WhatsApp, TikTok, Google Photos and YouTube, and operators are using fake reviews, download counts and step-by-step guides to trick victims. Once granted privileges, ClayRat can exfiltrate SMS, call logs and notifications, take front-camera photos, and even send messages or place calls while abusing Android's SMS handler role. Security firms report over 600 samples and coordinated disclosure to Google resulted in Play Protect protections.

Velociraptor Abused in Ransomware Attacks by Storm-2603

🔐 Cisco Talos confirmed ransomware operators abused Velociraptor, an open-source DFIR endpoint tool, to gain arbitrary command execution in August 2025 by deploying an outdated agent vulnerable to CVE-2025-6264. Talos links the activity with moderate confidence to Storm-2603 based on overlapping tooling and TTPs. Operators used the tool to stage lateral movement, deploy fileless PowerShell encryptors, and deliver multiple ransomware families, severely disrupting VMware ESXi and Windows servers.

AI-Powered Cyberattacks Escalate Against Ukraine in 2025

🔍 Ukraine's SSSCIP reported a sharp rise in AI-enabled cyber operations in H1 2025, documenting 3,018 incidents versus 2,575 in H2 2024. Analysts found evidence that attackers used AI not only to craft phishing lures but also to generate malware samples, including a PowerShell stealer identified as WRECKSTEEL. Multiple UAC clusters—such as UAC-0219, UAC-0218, and UAC-0226—deployed stealers and backdoors via booby-trapped archives, SVG attachments, and ClickFix-style tactics. The report also details zero-click exploitation of Roundcube and Zimbra flaws and widespread abuse of legitimate cloud and collaboration services for hosting and data exfiltration.

GitHub Copilot Chat prompt injection exposed secrets

🔐 GitHub Copilot Chat was tricked into leaking secrets from private repositories through hidden comments in pull requests, researchers found. Legit Security researcher Omer Mayraz reported a combined CSP bypass and remote prompt injection that used image rendering to exfiltrate AWS keys. GitHub mitigated the issue in August by disabling image rendering in Copilot Chat, but the case underscores risks when AI assistants access external tools and repository content.

Optical Mice Can Be Used to Eavesdrop on Conversations

🖱️ Researchers at the University of California, Irvine demonstrated a proof-of-concept called Mic-E-Mouse, showing that high-end optical mice can pick up desk-transmitted voice vibrations and be used to reconstruct nearby conversations. The attack can be executed on PC, Mac and Linux by non-privileged user-space programs, and Wiener and neural-network filtering was used to enhance muffled signals into intelligible speech. Practical limits include a quiet environment, thin desks (≈3 cm or less), mostly stationary mice and very high-DPI hardware; placing a rubber pad or mouse mat under the mouse prevents the leakage.

OpenAI Disrupts Malware Abuse by Russian, DPRK, China

🛡️ OpenAI said it disrupted three clusters that misused ChatGPT to assist malware development, including Russian-language actors refining a RAT and credential stealer, North Korean operators tied to Xeno RAT campaigns, and Chinese-linked accounts targeting semiconductor firms. The company also blocked accounts used for scams, influence operations, and surveillance assistance and said actors worked around direct refusals by composing building-block code. OpenAI emphasized that models often declined explicit malicious prompts and that many outputs were not inherently harmful on their own.

BatShadow Deploys Go-Based Vampire Bot Against Job Seekers

🔎 A Vietnam-linked group tracked as BatShadow is running a social-engineering campaign that lures job seekers and digital marketing professionals with faux job descriptions to deliver a previously undocumented Go-based malware, Vampire Bot. Attackers distribute ZIP archives containing decoy PDFs alongside malicious LNK or executable files that launch an embedded PowerShell script to fetch lure documents and remote-access tooling such as XtraViewer. The lure coerces victims into opening links in Microsoft Edge, triggering an automatic ZIP download that contains a deceptive executable padded to appear as a PDF; once executed, the Go binary profiles the host, exfiltrates data, captures screenshots, and maintains contact with a command-and-control server.

Gemini Trifecta: Prompt Injection Exposes New Attack Surface

🔒 Researchers at Tenable disclosed three distinct vulnerabilities in Gemini's Cloud Assist, Search personalization, and Browsing Tool. The flaws let attackers inject prompts via logs (for example by manipulating the HTTP User-Agent), poison search context through scripted history entries, and exfiltrate data by causing the Browser Tool to send sensitive content to an attacker-controlled server. Google has patched the issues, but Tenable and others warn this highlights the risks of granting agents too much autonomy without runtime guardrails.

XWorm Backdoor Returns with Ransomware and 35+ Plugins

🛡️ New variants of the XWorm backdoor (6.0, 6.4, 6.5) are being distributed via phishing campaigns after the original author, XCoder, abandoned the project. Multiple operators have adopted these builds, which now support more than 35 plugins enabling data theft, remote control, and a ransomware module that encrypts user files and drops HTML ransom notes. Trellix observed diverse droppers and recommends layered defenses including EDR, email/web protections, and network monitoring.