From The Hacker News · 40 stories
Australian Signals Directorate Disconnects 180,000 Computers from Amadey Malware Backend
The Australian Signals Directorate (ASD) gained access to an Amadey malware operator's backend and disconnected 180,000 infected computers globally. This action prevents the criminal from accessing or stealing data from these machines, though previously stolen credentials may still be at risk.
OpenAI Compaction Flaw Exposed Hidden Model Reasoning to Researchers
Security firm Xint discovered a flaw in OpenAI's compaction feature that allowed researchers to extract a model's hidden reasoning in plain text. This method bypassed protections OpenAI implemented in August against reasoning theft, but OpenAI confirmed and fixed the issue on October 23.
Critical Wasmtime Flaws Allow WebAssembly Sandbox Escapes, Affecting Multiple Versions
Two critical vulnerabilities in Wasmtime, a WebAssembly runtime, could allow malicious WebAssembly modules to escape their intended sandboxes. These flaws, rated 9.9 on the CVSS scale, primarily affect services that compile untrusted user-provided modules. Fixed versions were released on October 19 and 20, and users are advised to update immediately.
uCreator Android Spyware Steals Telegram 2FA and Clones WeChat Accounts
A rentable Android spyware kit named uCreator is taking over Telegram and WeChat accounts by embedding malicious code within modified versions of these applications. This spyware, identified as Baohuo by Doctor Web in 2025, primarily targets users who sideload these tampered apps or use the kit makers' container app, posing a risk to mobile wallet and messaging app users.
Ruby Fixes Three Regex Engine Bugs Addressing Four Critical GitLab RCE Flaws
Ruby released versions 4.0.9, 3.4.12, and 3.3.13 to patch three bugs in its Onigmo regular expression engine. These bugs were responsible for four critical remote code execution vulnerabilities in GitLab, which were patched over the past month. Any application that compiles regular expressions from attacker-controlled text is affected, not just GitLab.
Critical Flaw in Python's Pillow Library Allows Code Execution via Crafted Image Filenames
A critical vulnerability (CVE-2026-112900) in Python's Pillow image library allows remote code execution by exploiting how it handles EPS image filenames passed to Ghostscript. Attackers can craft filenames to execute commands on servers, impacting applications that process user-supplied images. The flaw has been patched in Pillow version 13.0.0, released on October 16.
Critical Flaw in x402 SDK Could Allow Malicious APIs to Drain AI Agent Token Balances
A critical vulnerability in the x402 SDK's default client allowed malicious APIs to request and receive an AI agent's full token balance due to a missing spending limit. This flaw, along with two others, was fixed in August 18 releases, but no public security advisories were issued, leaving developers unaware of the need to upgrade.
Two CVSS 9.0 RCE Flaws Disclosed in Linux KSMBD Before Kernel Fixes
Two remote code execution vulnerabilities (ZDI-CAN-33774 and ZDI-CAN-33549) with CVSS 9.0 ratings have been publicly disclosed in Linux KSMBD, the kernel's built-in SMB server, before official fixes are available in the mainline kernel. These flaws could allow attackers to execute arbitrary code on affected Linux servers, with one requiring no authentication.
New QPACK Flaws Allow Remote HTTP/3 Server Crashes and Memory Amplification
Security firm FoxIO disclosed vulnerabilities in HTTP/3 QPACK implementations that allow unauthenticated remote attackers to crash servers via denial-of-service attacks. These flaws enable memory amplification up to 4,000-fold by sending compressed HTTP headers that expand significantly upon unpacking, affecting multiple widely used HTTP/3 stacks.
China-Aligned Bronze Sandcastle Group Uses Sandboxie DLL Sideloading for Backdoors
Cybersecurity researchers identified Bronze Sandcastle, a China-aligned threat group active since mid-2025, which uses Sandboxie DLL sideloading to deploy backdoors. This technique exploits how Windows loads DLLs, allowing malicious code to run with the privileges of legitimate applications and evade detection.
State Bank of India Investigates Data Leak Affecting 3,897 Customer Records
State Bank of India (SBI) confirmed that 3,897 customer records containing personal information were found on the dark web, though the source and authenticity are still under investigation. The exposed data includes names, mobile numbers, dates of birth, and addresses, raising concerns about potential impersonation for affected customers.
Go Emergency Update Blocks Eight CA Keys Due to Undisclosed Security Concerns
Go versions 1.27.3 and 1.26.10 were released to block eight intermediate certificate authority keys due to "specific security concerns around their operation." This update causes Go programs to reject connections to sites and services using certificates issued under these keys, impacting certificate validation for Go applications.
Siemens Software Center Flaw Exploited in ClickFix Attacks to Deploy RAT
Siemens addressed a DLL sideloading vulnerability (CVE-2026-110936) in its Software Center, which attackers exploited in ClickFix campaigns to deploy remote access Trojans. Attackers used a signed Siemens application to load malicious DLLs, leading to arbitrary code execution.
Unpatched Broadcom Wi-Fi Flaw Allows Remote Code Execution on Routers
A security researcher disclosed an unpatched flaw in Broadcom's Wi-Fi Protected Setup (WPS) software that allows attackers within radio range to execute code as root on affected routers without authentication. The vulnerability, which affects millions of devices, stems from mishandling fragmented WPS messages, enabling memory corruption and remote code execution.
OpenCode Console Breach Exposed OpenAI, Anthropic, and Google API Keys
OpenCode reported a breach of its Console service on October 16, potentially exposing third-party AI provider keys for 1.4% of its Console users. Affected users have been advised to revoke and replace their OpenAI, Anthropic, and Google API keys, as some users have already reported unauthorized usage and associated costs.
Help TDS Linked to Hotejka Affiliate Program, Over 20,000 Cloudflare Pages Domains Host Tech Support Scams
A traffic distribution system (TDS) named Help TDS, active since 2017, is now linked to the Russian-language affiliate program Hotejka, which is suspected to be a rebrand of ROI777. This operation uses compromised websites, including over 20,000 Cloudflare Pages domains, to redirect users to tech support scams that trick victims into installing remote desktop tools.
NGINX Patches Five Vulnerabilities, Including Potential RCE and TLS Client-Cert Bypass
NGINX released versions 1.30.6 and 1.31.7 to address five vulnerabilities, including three memory bugs that could lead to crashes or remote code execution, a TLS flaw that might bypass client-certificate checks, and an HTTP/3 request smuggling issue. These flaws require specific NGINX configurations to be exploitable, and none are currently known to be exploited in the wild.
Next.js Releases Out-of-Band Patch for Critical RCE and DoS Vulnerabilities
Next.js released an out-of-band security update addressing two critical unauthenticated remote code execution (RCE) flaws and a high-severity denial-of-service (DoS) vulnerability. These fixes are crucial for users running Next.js 16 and 15 LTS releases to prevent potential server compromise and service disruption.
India Mandates Registered Caller Name Display on 4G and Newer Networks
The Indian government will require telecom service providers to display registered caller names for incoming calls on 4G and newer networks starting November 1, 2026. This measure aims to help users identify unknown callers and combat spam or fraudulent calls.
curl Patches HTTP/2 Upload Flaw Allowing Client-Side Code Execution
curl released version 8.23.0 to fix a use-after-free vulnerability (CVE-2026-92392) in its HTTP/2 upload handling that could lead to client-side memory corruption and potential code execution. This flaw affects curl and libcurl versions 8.12.0 through 8.22.0 and requires clients to upload a file to a malicious server.
Researchers Execute Code in Windows SmartScreen Enclave via Four-Byte Write and Timing Bug
Security researchers demonstrated code execution within a protected Windows SmartScreen enclave, bypassing its isolation mechanisms. This proof-of-concept attack, tracked as CVE-2026-83498, allowed code to run in a VBS enclave designed to guard sensitive data, even with system-level compromise. Microsoft patched the vulnerability in September 2026.
LG Smart TVs Vulnerable to Local Network Root Takeover via webOS Vulnerabilities
Three vulnerabilities in LG's webOS can be chained to allow an attacker on the same local network to gain root access to certain smart TVs without authentication. LG has released a fix for these vulnerabilities, which were demonstrated at TyphoonPWN 2026.
Two Critical Flaws in MarkLogic Server Allow Privilege Escalation to Admin Control
Progress Software disclosed 31 new flaws in its MarkLogic Server database, including two critical vulnerabilities that allow delegated managers to gain full administrator control. Fixes are available in MarkLogic 11.3.7 and 12.1.0, and users of affected versions should upgrade immediately.
GoBalance Flaw Allows .onion Address Hijacking via Tor-Format Key Recovery
A vulnerability in GoBalance, a tool used by dark-web sites for resilience, allows attackers to recover a site's .onion address private key using only public information. This flaw enables redirection of site visitors to attacker-controlled copies, impacting sites that store their master key in Tor's specific key format.
Japan Reports Sharp Increase in Web Data Leaks from Mobile API Abuse and Metabase Attacks
JPCERT/CC reported a surge in personal data leaks at Japanese organizations, attributing them to mobile app API abuse and exploitation of known software flaws, including in the Metabase BI tool. These incidents, separate from ransomware, have led to large-scale data compromises and are increasing, with 81 of 119 incidents this year occurring since July.
UAC-0099 Targets Ukrainian Government with New ASHVEIN RAT
The Russia-aligned threat actor UAC-0099 is using a new .NET infostealer and remote access trojan (RAT) called ASHVEIN in attacks against Ukrainian government personnel. ASHVEIN combines credential theft, surveillance, and remote-control capabilities, hiding commands within invisible HTML elements.
Critical LMCache Flaw Allows Unauthenticated Remote Code Execution
A critical vulnerability in LMCache, an open-source tool for LLM servers, allows unauthenticated remote code execution due to insecure deserialization in its multiprocess mode. The flaw, tracked as CVE-2026-105192, affects versions 0.3.9 through 0.5.5 and has no available patch, posing a risk to LLM deployments configured with routable addresses.
Phishing Platform Impersonates AI Chatbot Ad Portals to Steal Credentials and MFA Codes
Cybersecurity researchers uncovered a human-operated phishing platform that mimics advertising products for AI chatbots like ChatGPT, Gemini, and Claude. The platform uses browser-in-the-browser (BitB) attacks to capture user credentials and multi-factor authentication codes by displaying fake login windows.
LibreOffice and OpenOffice Flaws Allow Code Execution via Malicious Spreadsheets
Security researchers discovered vulnerabilities in LibreOffice and Apache OpenOffice that allow malicious spreadsheets to execute code without macro warnings when Java support is enabled. LibreOffice has patched its flaw (CVE-2026-63277) in recent updates, while Apache OpenOffice (CVE-2026-59265) remains vulnerable, with a fix expected in version 4.1.17. This impacts users of both office suites, particularly those who open untrusted documents with Java enabled.
China-Aligned TA419 Group Targets U.S. AI Policy Experts with Microsoft AitM Phishing
A China-nexus cyber espionage group, TA419, is conducting credential phishing campaigns against U.S. AI policy experts in think tanks, universities, and legal organizations. The attacks use sophisticated adversary-in-the-middle (AitM) techniques to gather intelligence on U.S. AI policy and regulatory developments. This activity indicates ongoing strategic competition and intelligence gathering efforts related to AI between the U.S. and China.
MI5 Warns 100+ UK Academics Aided Chinese Intelligence Research
MI5 issued an alert stating that China's Ministry of State Security (MSS) funded research involving over 100 UK-linked academics through a front company, CGTRI. This activity supports MSS espionage and poses a threat to UK national security, prompting warnings about potential prosecution under the National Security Act 2023.
Android 17 Advanced Protection Limits Accessibility Services to Verified Tools
Google will restrict Android's accessibility services to verified Accessibility Tools applications when Advanced Protection is enabled in Android 17. This change aims to block a major attack pathway exploited by malicious applications for financial fraud and data theft.
OFAC Sanctions Tren de Aragua for ATM Jackpotting; Threat Actors Use Blockchains for Malware
The U.S. Treasury's OFAC sanctioned 10 individuals involved in a Tren de Aragua ATM jackpotting scheme that stole over $40 million from U.S. financial institutions, with proceeds laundered via cryptocurrency. Separately, cyber threat actors are using public blockchains, a technique called EtherHiding or Blockchain Dead Drops, to conceal malware instructions, making takedowns difficult.
WordPress Backdoor Uses Multiple Persistence Mechanisms to Self-Rebuild After Cleanup
Cybersecurity researchers have identified a sophisticated WordPress backdoor, codenamed SC, that employs multiple persistence mechanisms across files, the database, and shared memory to rebuild itself even after cleanup attempts. This malware creates a circular system where each component can restore others, making it difficult to remove completely.
Attackers Abuse MSP360 to Deploy ScreenConnect in Dual-RMM Phishing Attacks
Microsoft has warned of phishing campaigns distributing MSP360 Remote Monitoring and Management (RMM) software to gain initial access to systems. Attackers then use this foothold to install ConnectWise ScreenConnect, establishing redundant remote access channels for further malicious activity.
OpenSSL Fixes High-Severity DTLS Flaw That Can Leak Heap Memory or Crash Programs
OpenSSL released fixes for a high-severity DTLS flaw, CVE-2026-84782, that can leak heap memory or crash programs during handshake message resends. The vulnerability affects software using OpenSSL for DTLS and is patched in versions 4.0.3, 3.6.5, 3.5.9, and 3.4.8, with premium support required for older branches.
French Tax Data Stolen Using Staff Passwords, Undetected for Seven Weeks
An attacker stole tax data from hundreds of thousands of French taxpayers and businesses using compromised staff passwords, remaining undetected for seven weeks. The breach was attributed to weak login protection, poor network segmentation, and monitoring gaps within the tax administration's systems.
Official MCP Python SDK Flaw Allows Malicious Servers to Steal OAuth Credentials
A security flaw in the official MCP Python SDK allowed malicious servers to steal OAuth credentials, including client secrets and authorization codes, from applications. This vulnerability enabled attackers to obtain valid access tokens with the application's permissions, impacting applications using specific OAuth providers over HTTP.
RatHat Android Malware Console Uses Google Gemini to Prioritize Victims
The RatHat Android banking trojan's latest console version integrates Google's Gemini AI to analyze victim text messages and estimate bank balances, sorting victims into high-value and mid-value groups. This allows operators to prioritize their efforts, indicating a shift in malware-as-a-service tactics towards more efficient victim targeting.
PamStealer macOS Malware Updates Payload Decryption and Persistence Mechanisms
A new version of PamStealer macOS malware now uses server-side decryption for its main payload and employs a different lure, advertising a non-existent cryptocurrency wallet service. The updated malware also implements four redundant persistence methods, including suppressing macOS notifications for new background login items.