• A high-severity vulnerability was identified in LangChainGo, the Go implementation of the popular LLM orchestration framework LangChain. 

    Tracked as CVE-2025-9556, this flaw allows unauthenticated attackers to perform arbitrary file reads through maliciously crafted prompt templates, effectively exposing sensitive server files without requiring direct system access.

    Key Takeaways
    1. CVE-2025-9556, Jinja2 prompt injection enables arbitrary file reads.
    2. Requires only prompt access, exposing shared deployments.
    3. Fixed with RenderTemplateFS or NewSecureTemplate.

    Server-Side Template Injection

    LangChainGo relies on the Gonja template engine, a Go port of Python’s Jinja2, to parse and render dynamic prompts. 

    The CERT Coordination Center and the Software Engineering Institute report that Gonja’s compatibility with Jinja2 directives such as {% include %}, {% from %}, and {% extends %} enables reusable templates but also introduces dangerous file-system interactions when untrusted content is rendered. 

    An attacker submits a prompt containing a payload like:

    Critical LangChainGo Vulnerability

    This can force LangChainGo into loading and returning the contents. Because Gonja processes Jinja2 syntax natively, advanced template constructs such as nested statements or custom macros can be used to traverse directories or chain multiple file reads in a single injection string. 

    In LLM chat environments powered by LangChainGo, the only prerequisite is access to the prompt submission interface, making exploitation trivial for remote threat actors.

    Risk FactorsDetails
    Affected ProductsLangChainGo < 0.18.2
    ImpactArbitrary file read; data breach
    Exploit PrerequisitesAccess to LLM prompt interface
    CVSS 3.1 Score9.8 (Critical)

    Mitigations

    The vulnerability compromises confidentiality and undermines the core trust model of LLM-based systems. Attackers can harvest SSH keys, environment files, API credentials, or other proprietary data stored on the server. 

    Once in possession of these files, adversaries may elevate privileges, pivot laterally, or exfiltrate intellectual property. The risk is magnified in multi-tenant deployments where one malicious user could access the filesystem resources of another tenant’s instance.

    To remediate, maintainers have released a patch that introduces a secure RenderTemplateFS function, which enforces a whitelist of permissible template paths and disables arbitrary filesystem access by default. 

    The update also hardens template parsing routines to sanitize or reject any prompt containing Jinja2 file-inclusion directives. Operators of LangChainGo should immediately upgrade to version 0.18.2 or later and audit their prompt-handling code for any custom template instantiation using NewTemplate(), replacing it with the patched NewSecureTemplate API().

    Find this Story Interesting! Follow us on Google NewsLinkedIn, and X to Get More Instant Updates.

    The post Critical LangChainGo Vulnerability Let Attackers Access Sensitive Files by Injecting Malicious Prompts appeared first on Cyber Security News.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • Security researchers have uncovered a sophisticated malware campaign spanning seven years, where threat actors behind AppSuite-PDF and PDF Editor applications systematically abused code-signing certificates to legitimize their malicious software.

    The actors, tracked under the malware family name BaoLoader, have utilized at least 26 code-signing certificates obtained through fraudulent business registrations, primarily targeting users seeking PDF editing tools and productivity applications.

    The campaign demonstrates a calculated approach to certificate authority manipulation, with threat actors establishing legitimate businesses across multiple jurisdictions including Panama, Malaysia, and the United States.

    PDF Editor (Source – Expel)

    These entities served as fronts for obtaining code-signing certificates from major certificate authorities including SSL.com, GlobalSign, DigiCert, and Sectigo.

    The malware has been distributed under various guises, including AppSuite-PDF, PDF Editor, ManualFinder, PDFTools, PDFProSuite, and OneStart, often masquerading as potentially unwanted programs (PUPs) while harboring backdoor capabilities.

    What sets this campaign apart from typical certificate abuse scenarios is the actors’ consistent pattern of obtaining multiple certificates for identical company names from different certificate authorities.

    Expel researchers identified this unusual behavior while analyzing the CertCentral.org database, noting that among over 1,500 documented organizations with abused certificates, these actors were unique in their geographic certificate patterns and systematic approach to legitimacy simulation.

    The threat actors’ methodology extends beyond simple certificate procurement to include sophisticated distribution mechanisms.

    Files have been uploaded to platforms like VirusTotal under numerous deceptive names, with single executables appearing as “ZoomSetup,” “WinRarSetup,” “MinecraftSetup,” and various PDF-related applications.

    This multi-naming strategy indicates deliberate attempts to maximize infection vectors by appealing to diverse user interests and needs.

    Certificate Authority Exploitation and Business Registration Patterns

    The technical infrastructure behind BaoLoader reveals meticulous planning in certificate acquisition strategies.

    The actors established companies with media-focused naming conventions, including GLINT SOFTWARE SDN. BHD., ECHO INFINI SDN. BHD., Summit Nexus Holdings LLC, Apollo Technologies Inc., and Caerus Media LLC.

    Each entity was registered with legitimate business documentation, enabling the actors to pass initial certificate authority verification processes.

    The Digicert-issued certificate for Eclipse Media Inc. uses the same RDN number as the same company name certificates issued by Sectigo, GlobalSign, and SSL.com (Source – Expel)

    Analysis of the certificate metadata reveals consistent business serial numbers across multiple certificate authorities for identical company names.

    For instance, Eclipse Media Inc. certificates were issued by GlobalSign, SSL.com, Sectigo, and DigiCert, all containing matching business registration identifiers.

    This approach allowed the actors to maintain operational continuity when individual certificates faced revocation, seamlessly transitioning between certificate authorities while maintaining the same organizational identity.

    The malware’s persistence mechanisms include PowerShell execution designed to load Web Companion components, executing commands that bypass execution policies and load assemblies from encrypted files.

    The campaign’s evolution from simple adware distribution to backdoor deployment represents a concerning escalation in threat actor capabilities, demonstrating how certificate abuse can provide extended operational longevity for malicious campaigns targeting productivity software users.

    Boost your SOC and help your team protect your business with free top-notch threat intelligence: Request TI Lookup Premium Trial.

    The post Actors Behind AppSuite-PDF and PDF Editor Used 26 Code-Signing Certificates to Make Software Appear Legitimate appeared first on Cyber Security News.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • The Kimsuky APT group has begun leveraging generative AI ChatGPT to craft deepfake South Korean military agency ID cards. Phishing lures deliver batch files and AutoIt scripts designed to evade anti-virus scanning through sophisticated obfuscation. Organizations must deploy endpoint detection and response (EDR) solutions to unmask hidden scripts and secure endpoints. On July 17, 2025, […]

    The post Hackers using generative AI “ChatGPT” to evade anti-virus defenses appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • A critical permission misconfiguration in the IBM QRadar Security Information and Event Management (SIEM) platform could allow local privileged users to manipulate configuration files without authorization. 

    Tracked as CVE-2025-0164, the flaw stems from improper permission assignment and carries a CVSS 3.1 base score of 2.3 (AV:L/AC:L/PR:H/UI:N/S:U/C:L/I:N/A:N).

    Key Takeaways
    1. CVE-2025-0164 in QRadar SIEM v7.5–7.5.0 UP13 IF01 lets privileged locals alter config files.
    2. Vulnerability stems from CWE-732 (Incorrect Permission Assignment for Critical Resource).
    3. Apply UP13 IF02, limit admin access, and watch /opt/qradar/conf.

    Incorrect Permission Assignment Flaw

    The vulnerability arises from incorrect permission assignment for critical resources (CWE-732), which fails to enforce appropriate access controls on configuration directories and files within QRadar SIEM installations running versions 7.5 through 7.5.0 UP13 IF01. 

    A local user with existing high-level privileges, such as a system administrator or support engineer, can exploit the flawed file system permissions to alter key configuration parameters, modify logging policies, or disable detection rules.

    Attackers could script automated modifications by invoking shell commands against protected paths.

    These unauthorized changes may persist until remedied by manual intervention, and could frustrate incident response efforts by masking malicious activity in audit logs or allowing further unauthorized actions without detection.

    Risk FactorsDetails
    Affected ProductsIBM QRadar SIEM 7.5–7.5.0 UP13 IF01
    ImpactUnauthorized modifications to config files, disabling rules or altering logging policies
    Exploit PrerequisitesLocal privileged user access
    CVSS 3.1 Score2.3 (Low)

    Mitigations

    To remediate CVE-2025-0164, IBM has released QRadar 7.5.0 UP13 IF02, which corrects file and directory permissions to restrict write access exclusively to the QRadar service account. 

    Administrators should apply the interim fix immediately on affected systems by downloading the update from IBM Fix Central. 

    The applicable fix can be retrieved using fix ID 7.5.0-QRADAR-QRSIEM-20250904123850INT. No workaround exists for environments where privileged users are permitted shell-level access. 

    As a precaution, organizations should restrict local administrative privileges to trusted personnel only and monitor filesystem changes in /opt/qradar/conf.

    Maintaining robust access controls and timely patching remains essential to preserving the integrity of security monitoring infrastructures.

    Find this Story Interesting! Follow us on Google NewsLinkedIn, and X to Get More Instant Updates.

    The post IBM QRadar SIEM Vulnerability Let Attackers Perform Unauthorized Actions appeared first on Cyber Security News.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • Countries with most cyberattacks stopped highlighting global cyber defense efforts, including key regions in Asia-Pacific and North America.  BlackNevas has released a comprehensive attack strategy spanning three major regions, with the Asia-Pacific area bearing the heaviest burden of attacks at 50% of total operations. The group’s primary targets in this region include major economies such as […]

    The post BlackNevas Ransomware Encrypts Files, Exfiltrates Corporate Data appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • The Great Firewall of China (GFW) suffered its largest-ever internal data breach. More than 500 GB of sensitive material—including source code, work logs, configuration files, and internal communications—was exfiltrated and published online. 

    The breach stems from Geedge Networks and the MESA Lab at the Institute of Information Engineering, Chinese Academy of Sciences. 

    The leaked archive reveals the GFW’s R&D workflows, deployment pipelines, and surveillance modules used across Xinjiang, Jiangsu, and Fujian provinces, as well as export agreements under China’s “Belt and Road” framework to Myanmar, Pakistan, Ethiopia, Kazakhstan, and other undisclosed nations. 

    Key Takeaways
    1. 500 GB+ of GFW internals leaked, exposing DPI engines and surveillance code.
    2. 600 GB archive available via BitTorrent/HTTPS; key file repo.tar.
    3. Use isolated VMs, verify hashes, and don’t run unvetted binaries.

    Analysts warn that exposed internals such as the DPI engine, packet filtering rules, and update signing certificates will enable both evasion techniques and deep insight into censorship tactics.

    Key file manifests include:

    Great Firewall of China Sensitive Data Leaked

    For unpacking, use tar -xvf repo.tar on a secure host.

    Operational Security Protocols

    Given the leak’s sensitivity, downloading or analyzing these datasets poses significant security and legal risks. 

    Files may contain proprietary encryption keys, surveillance configuration scripts, or malware-laden installers, potentially triggering remote monitoring or defensive countermeasures. 

    Researchers should adopt stringent operational security protocols:

    • Analyze within an isolated virtual machine or air-gapped sandbox running minimal services. 
    • Employ network-level packet captures and snapshot-based rollback to detect and contain malicious payloads. Always verify file hashes (SHA-256 sums provided in mirror/filelist.txt) before extraction. 
    • Avoid executing binaries or running build scripts without code review. Many artifacts include custom kernel modules for deep packet inspection that could compromise host integrity. 

    Obfuscation techniques discovered in mesalab_git.tar.zst use polymorphic C code and encrypted configuration blocks; reverse-engineering without safe-lab instrumentation may trigger anti-debugging routines. 

    Researchers are encouraged to coordinate with trusted malware analysis platforms and disclose findings responsibly. This unprecedented leak grants the security community an unusual view behind the GFW’s opaque infrastructure.

    Find this Story Interesting! Follow us on Google NewsLinkedIn, and X to Get More Instant Updates.

    The post Great Firewall of China’s Sensitive Data of Over 500GB+ Leaked Online appeared first on Cyber Security News.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • DarkCloud Stealer has recently emerged as a potent threat targeting financial organizations through convincing phishing campaigns. Adversaries employ weaponized RAR attachments masquerading as legitimate documents to deliver a multi-stage JavaScript-based payload.

    Upon opening the archive, victims execute a VBE script that leverages Windows Script Host to initiate a PowerShell downloader hidden in innocuous-seeming image files.

    This initial access vector exploits users’ trust in routine financial correspondence, triggering an automated chain of decoding and decryption steps designed to evade conventional security controls.

    In early September 2025, security teams observed a dramatic uptick in malicious RAR attachments sent to corporate email accounts within the banking sector.

    CyberProof analysts identified that the archive named “Proof of Payment.rar” contains a VBE script which, when executed, calls PowerShell to download an embedded JPG file named universe-1733359315202-8750.jpg.

    Device timeline showing the download activty from user (Source – CyberProof)

    The stealer’s loader is concealed within this image, and the decoding routine extracts the .NET DLL module directly from image pixel data.

    CyberProof researchers noted that the PowerShell script rigorously checks memory offsets to locate a distinct BMP header pattern before carving out the loader DLL.

    The following snippet illustrates the core loop used for scanning the downloaded image bytes:-

    for ($i=0; $i -lt $data.Length - $header. Length; $i++) {
        $match = $true
        for ($j=0; $j -lt $header.Length; $j++) {
            if ($data[$i + $j] -ne $header[$j]) { $match = $false; break }
        }
        if ($match) { $offset = $i; break }
    }

    Once the DLL is reconstructed in memory, the script invokes [Reflection.Assembly]::Load() to execute the loader without ever touching disk.

    Persistence and Credential Theft

    After loading into memory, DarkCloud Stealer establishes persistence by copying a JavaScript payload to the Windows Run registry key under a disguised filename (M3hd0pf.exe masquerading as MSBuild.exe), ensuring execution on every user login.

    The stealer then injects into legitimate processes like MSBuild.exe and mtstocom.exe using process hollowing techniques, enabling it to siphon saved credentials from browser databases such as Chrome’s Login Data.

    Alerts from endpoint detection platforms confirm DPAPI access events and memory mapping into browser processes, revealing attempts to decrypt stored passwords directly in memory.

    Stolen data being sent to remote IPs (Source – CyberProof)

    Finally, stolen data is staged in user directories and exfiltrated via FTP and HTTP channels to dynamic domain clusters (.shop, .xyz), complicating network-based detection.

    Financial institutions are urged to monitor for anomalous VBE/VBS execution, unexpected registry Run key modifications, and JavaScript files in public download folders to rapidly detect and disrupt this insidious campaign.

    Boost your SOC and help your team protect your business with free top-notch threat intelligence: Request TI Lookup Premium Trial.

    The post DarkCloud Stealer Attacking Financial Companies With Weaponized RAR Attachments appeared first on Cyber Security News.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • A recently discovered flaw in LangChainGo, the Go implementation of the LangChain framework for large language models, permits attackers to read arbitrary files on a server by injecting malicious prompt templates. Tracked as CVE-2025-9556, this vulnerability arises from the use of the Gonja template engine, which supports Jinja2 syntax and can be manipulated to perform […]

    The post LangChainGo Vulnerability Allows Malicious Prompt Injection to Access Sensitive Data appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • The Federal Bureau of Investigation (FBI) has released a detailed flash advisory disclosing indicators of compromise (IOCs) and tactics used by two cybercrime groups—UNC6040 and UNC6395—to breach Salesforce customer environments and siphon sensitive data. Coordinated with the Department of Homeland Security Cybersecurity and Infrastructure Security Agency (DHS/CISA), the bulletin aims to equip security teams and […]

    The post FBI Releases IOCs on Cyber Threats Exploiting Salesforce for Data Theft appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

  • PEARL HARBOR, Hawaii—Robot boats. Counter-drone systems. A prototype Army landing craft. A million dollars in cash prizes. It’s all part of the Army’s effort to overcome logistics challenges in the Indo-Pacific and get new technology in the hands of soldiers more quickly. 

    Experimentation for the Army’s xTech contest started Sept. 9 with sunshirt-clad representatives from HavocAI wheeling small gray autonomous boats down a concrete ramp on Ford Island into the cerulean waters of Pearl Harbor. Soon, the boats were zipping through the harbor, protecting a Maneuver Support Vessel (Light) from an unnamed adversary. A total of 10 companies tested their tech as part of the contest. 

    On Thursday, three winners will take home $35,000 cash prizes, along with feedback from soldiers and technical experts that will help as they pursue a follow-on contract. 

    “We partnered with [U.S. Army Pacific] to identify some of their capability gaps, and invited industry to solve some of those [challenges]. Our goal is really to bring in new industry and offer them a friendly front door to the Army where they can propose their solutions, but also get that direct feedback from stakeholders,” said Jessica Stillman, project manager for the Army’s xTech program

    The 10 competing companies were winnowed from a pool of about 130 that submitted white papers after a call for solutions, Stillman said. About 44 percent of the companies that apply for these contests are new to the Defense Department, and the service tries “to make it as easy as possible” for them throughout the process.

    “A lot of times we keep the problem statements pretty broad, because sometimes industry can tell us what we need,” she said. “You know, we don’t always know what industry is doing, and that’s the value of using our program, because it allows you to do that market research and understand what industry is developing, and can that apply to what the Army needs. It gives the Army the opportunity to bring in new innovation, and new companies into the pipeline, which is really great.”

    Though xTech has run nearly 50 contests so far, this is the first that involves the competitors running scenarios in an experimentation environment, Stillman said. 

    The Pearl Harbor experiments deployed drone boats and counter-drone tech to protect a high-value asset, while experiments at inland Schofield Barracks focused on unexploded ordnance and electronic warfare. 

    The role of “high-value asset” for the exercise was played by a prototype logistics vessel called MSV(L)-01, carrying two Strykers on its reinforced aluminum deck. Vessel Master Chief Warrant Officer 2 Matthew Davis said the crew is testing the vessel to inform low-rate initial production. 

    “Today was kind of a two-birds, one-stone” situation, Davis said: testing the requirements of the vessel along with serving as an asset to be defended by the robot boats. 

    The MSV(L) was built as a replacement for the Vietnam-era Landing Craft Mechanized, for use in the Indo-Pacific, Davis said. 

    “This vessel offers an increased capability and capacity to what we already have,” he said. “The draft is shallower, so fully laden is 82 short tons… With that payload on board, we can still make 23 knots with only five and a half feet of draft,” so even loaded up, it can go straight to the beach and set its ramp on the sand, while other Army vessels require a port. 

    While this iteration of the contest is the first to include experiments, Stillman said the xTech office has three others lined up in the next three months with different commands. 

    “One of the things that senior leaders are focusing on right now is getting the technology into the hands of the soldiers,” she said. “This is the first step.” 

    ]]>

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶

    ¶¶¶¶¶