A sophisticated new cyberattack targeting Android devices in South Korea has been uncovered, leveraging Google’s asset-tracking feature, Find Hub, to remotely wipe sensitive user data. Threat actors disguised as psychological counselors and North Korean human rights activists have distributed malware masquerading as stress-relief programs, marking a significant escalation in state-sponsored attacks linked to the notorious […]
A critical security flaw has been discovered in the widely used npm package expr-eval, potentially exposing AI and natural language processing applications to remote code execution attacks.
The vulnerability, tracked as CVE-2025-12735, allows attackers to execute arbitrary system commands through maliciously crafted input.
The expr-eval library is a JavaScript tool designed to parse and evaluate mathematical expressions safely, serving as a more secure alternative to JavaScript’s native eval() function.
With over 250 dependent packages, including oplangchain, a JavaScript implementation of the popular LangChain framework, this vulnerability has significant implications for the AI and NLP ecosystem.
NPM LibraryVulnerability
Carnegie Mellon University researchers discovered that attackers can define arbitrary functions within the parser’s context object, enabling the injection of malicious code that executes system-level commands.
This vulnerability achieves Total Technical Impact under the SSVC framework, meaning adversaries gain complete control over affected software behavior and can access all system information.
CVE ID
Affected Package
Vulnerability Type
Patched Version
CVE-2025-12735
expr-eval, expr-eval-fork
Remote Code Execution
expr-eval-fork v3.0.0
The flaw is particularly dangerous for generative AI systems and NLP applications. These systems often run in server environments with access to sensitive local resources and process user-supplied mathematical expressions.
Developers using expr-eval or expr-eval-fork should take immediate action by upgrading to the expr-eval-fork version 3.0.0, which includes comprehensive security patches.
The update introduces an allowlist of safe functions, mandatory registration for custom functions, and enhanced test cases to enforce security constraints.
The vulnerability was responsibly disclosed by security researcher Jangwoo Choe (UKO) and patched through GitHub Pull Request #288.
Organizations can use npm audit to automatically detect this vulnerability in their projects through the GitHub Security Advisory GHSA-jc85-fpwf-qm7x.
A critical remote code execution vulnerability has been discovered in LangGraph’s checkpoint serialization system.
The flaw CVE-2025-64439 affects versions of langgraph-checkpoint before 3.0. It allows attackers to execute arbitrary Python code when untrusted data is deserialized.
The vulnerability resides in LangGraph’s JsonPlusSerializer, the default serialization protocol used for checkpoint persistence.
When the serializer encounters illegal Unicode surrogate values during msgpack serialization, it automatically falls back to JSON mode.
LangGraph Vulnerability
In this fallback state, the deserializer supports a constructor-style format that reconstructs custom objects during loading.
The flaw carries a CVSS score of 8.5, classified as “High” severity. It falls under CWE-502, which covers deserialization of untrusted data vulnerabilities.
An attacker can exploit this mechanism to inject malicious payloads that execute arbitrary functions when deserialized.
Attribute
Details
CVE ID
CVE-2025-64439
Vulnerability Type
Remote Code Execution (RCE)
Component
LangGraph JsonPlusSerializer
Affected Versions
langgraph-checkpoint < 3.0
Patched Versions
langgraph-checkpoint >= 3.0
Severity
High (7.5 CVSS v4)
Attack Vector
Network
The vulnerability primarily affects applications that meet two conditions: They allow untrusted or user-supplied data to be persisted into checkpoints, and they use the default JsonPlusSerializer (or explicitly instantiate it).
Applications processing only trusted data face significantly lower practical risk. LangGraph-API deployments running version 0.5 or later are unaffected.
Attackers can weaponize this vulnerability through a carefully crafted proof-of-concept. That injects a malicious payload structured with specific JSON elements: an “lc” value of 2, a “type” set to “constructor,” and an “id” field pointing to dangerous system functions like “os.system.”
When the compromised checkpoint is loaded, the deserialization process automatically executes the embedded command.
LangChain has released langgraph-checkpoint version 3.0 to address this issue.
The fix implements an allow-list for constructor deserialization, restricting permissible paths to explicitly approved module and class combinations.
Additionally, saving payloads in JSON format has been deprecated to eliminate this unsafe fallback mechanism. Users should upgrade immediately to version 3.0 or later.
The update maintains full compatibility with langgraph 0.3. It requires no code modifications or import changes, making the patching process straightforward and low-risk for most deployments.
A sophisticated supply chain attack has compromised critical industrial control systems through nine malicious NuGet packages designed to inject time-delayed destructive payloads into database operations and manufacturing environments. Socket’s Threat Research Team identified these weapons of code, published under the alias shanhai666 between 2023 and 2024, which have collectively accumulated 9,488 downloads before being reported […]
The infamous Meow attack, which devastated unsecured databases since 2020, has resurfaced with renewed force through MAD-CAT (Meow Attack Data Corruption Automation Tool). This custom-built adversarial simulation tool demonstrates how easily attackers can corrupt data across multiple database platforms simultaneously, highlighting a critical vulnerability that continues to plague modern infrastructure. The Evolution of Meow While Meow attack […]
Cyber threats didn’t slow down last week—and attackers are getting smarter. We’re seeing malware hidden in virtual machines, side-channel leaks exposing AI chats, and spyware quietly targeting Android devices in the wild.
But that’s just the surface. From sleeper logic bombs to a fresh alliance between major threat groups, this week’s roundup highlights a clear shift: cybercrime is evolving fast
A critical remote code execution vulnerability has been discovered in the widely used JavaScript library expr-eval, affecting thousands of projects that rely on it for mathematical expression evaluation and natural language processing. The vulnerability, tracked as CVE-2025-12735, poses significant risks to server environments and to AI-powered applications that process user input. Identifier Value CVE ID […]
Intel has filed a federal lawsuit against a former employee accused of downloading thousands of classified documents shortly after being terminated, raising serious concerns about corporate data security and insider threats.
Jinfeng Luo, a software developer who has worked at Intel since 2014, was based in Seattle when the company notified him of his pending dismissal on July 7.
His employment officially ended on July 31 as part of Intel’s massive workforce reduction effort that saw over 15,000 employees laid off worldwide during the summer restructuring.
The Alleged Theft
According to Intel’s lawsuit filed in Washington federal court, Luo attempted to download files from his work laptop to an external hard drive on July 23. However, the company’s security controls blocked the transfer.
Luo allegedly connected a different storage device five days later and downloaded about 18,000 files. Many of these files were marked as “Top Secret” by Intel.
The unauthorized file transfer immediately triggered an internal investigation. Intel spent months attempting to contact Luo at his Seattle residence and two other addresses linked to him, including one in Portland, but the engineer could not be reached.
Unable to locate Luo, Intel pursued legal action seeking substantial compensation. The lawsuit demands at least $250,000 in damages, attorney fees, and a court injunction prohibiting Luo from disclosing any of the stolen confidential information to unauthorized parties.
Neither Intel nor Luo has provided public comments about the case. The tech giant declined to discuss details when contacted by media outlets, and Luo’s current whereabouts remain unknown.
Legal observers, OregonLive first reported on Intel’s lawsuit earlier this week through specialized law and employment news platforms.
The incident highlights significant vulnerabilities in protecting sensitive corporate data during layoffs.
As companies implement cost-cutting measures affecting thousands of employees, insider threats pose escalating risks to intellectual property and classified information.
Intel’s case demonstrates how quickly departing employees can access and remove sensitive materials before security measures can be fully implemented.
According to the new Browser Security Report 2025, security leaders are discovering that most identity, SaaS, and AI-related risks converge in a single place, the user’s browser. Yet traditional controls like DLP, EDR, and SSE still operate one layer too low.
What’s emerging isn’t just a blindspot. It’s a parallel threat surface: unmanaged extensions acting like supply chain implants, GenAI
Threat actors are systematically compromising Outlook and Google mailboxes with alarming success, leveraging sophisticated techniques that sidestep traditional email defenses entirely. According to VIPRE’s Q3 2025 Email Threat Report, over 90% of phishing attacks specifically target these two dominant email ecosystems, representing a calculated strategic shift by attackers seeking to maximize impact while minimizing operational […]
