Blahalla

www.bleepingcomputer.com · 4 months ago

A supply chain attack has compromised the popular LiteLLM Python package on PyPI, with malicious versions 1.82.7 and 1.82.8 containing an infostealer. The hacking group TeamPCP, previously linked to the Aqua Security breach, claimed responsibility for the attack. The malware harvests a wide range of credentials, including SSH keys, cloud tokens, and cryptocurrency wallet data, and attempts lateral movement within Kubernetes clusters, sending stolen data to models.litellm[.]cloud. Organizations using LiteLLM should immediately rotate credentials and check for persistence artifacts.

Meta Ordered to Pay $375 Million Over Child Safety Violations

www.nytimes.com · 4 months ago

NULL

Threat Brief: Recruiting Scheme Impersonating Palo Alto Networks Talent Acquisition Team

unit42.paloaltonetworks.com · 4 months ago

Attackers are impersonating Palo Alto Networks talent acquisition staff in sophisticated phishing campaigns targeting senior-level professionals. They use scraped LinkedIn data to send personalized emails, falsely claiming a candidate's resume doesn't meet applicant tracking system requirements and offering paid services like "executive ATS alignment" for $400 to $800. Victims are advised to verify sender domains, request official platforms, and treat payment requests as red flags. If targeted, individuals should cease communication, report the incident, and secure their accounts.

Checkmarx KICS Code Scanner Targeted in Widening Supply Chain Hit

www.darkreading.com · 4 months ago

Several open-source projects have recently been targeted in a series of supply chain attacks attributed to the threat actor TeamPCP. Attackers compromised the KICS GitHub Action, Checkmarx VS Code plugins, and poisoned versions of the Trivy security scanner and Litellm Python package. The malicious code, an infostealer, steals credentials including SSH keys, cloud credentials, and API tokens. Security experts warn organizations to immediately rotate all credentials and maintain a real-time inventory of compromised secrets to mitigate the risk.

How AI Coding Tools Crushed the Endpoint Security Fortress

www.darkreading.com · 4 months ago

New research indicates artificial intelligence coding tools are undermining endpoint security measures. These tools, including Anthropic’s Claude Code, OpenAI’s Codex, and Google’s Gemini, require broad access to developer endpoints, effectively bypassing traditional security layers. Researchers have identified vulnerabilities such as CVE-2025-59536 in Claude Code allowing malicious code execution and CVE-2025-61260 in OpenAI Codex CLI enabling code injection. To mitigate these risks, organizations should audit AI usage, isolate coding tools, and treat configuration files as executable code.

FCC bans new routers made outside the USA over security risks

www.bleepingcomputer.com · 4 months ago

The Federal Communications Commission has banned the sale of new consumer routers manufactured outside the United States due to national security risks. A recent assessment found these foreign-made routers could disrupt the U.S. economy and critical infrastructure, and have been linked to attacks by groups like Volt, Flax, and Salt Typhoon. While current routers remain available, accessing new models may become more difficult and costly, and manufacturers can seek approval by disclosing detailed corporate, manufacturing, and supply chain information, including a plan to move manufacturing to the United States. Certain routers used by the Department of War and Homeland Security are exempt.

Corelan: Debugging - WinDBG & WinDBGX Fundamentals -

www.corelan.be · 4 months ago

false

DoE Publishes 5-Year Energy Security Plan

www.securityweek.com · 4 months ago

false

US Bans New Foreign-Made Routers, Citing ‘Unacceptable’ Security Risks

www.techrepublic.com · 4 months ago

false

How a Poisoned Security Scanner Became the Key to Backdooring LiteLLM

snyk.io · 4 months ago

A supply chain attack compromised versions 1.82.7 and 1.82.8 of the "litellm" Python package, resulting in malicious code being injected. The attackers exploited a vulnerability in Trivy, CVE-2024-3712, to gain access to the maintainer’s credentials and use a compromised GitHub Action. This resulted in credential harvesting, data exfiltration, and the establishment of a persistent backdoor within Kubernetes. The incident underscores the vulnerabilities of CI/CD pipelines and the importance of robust security measures.

Why Agentic AI Systems Need Better Governance – Lessons from OpenClaw

www.securityweek.com · 4 months ago

false

TeamPCP Backdoors LiteLLM Versions 1.82.7–1.82.8 via Trivy CI/CD Compromise

thehackernews.com · 4 months ago

The threat actor TeamPCP has compromised the litellm Python package, pushing versions 1.82.7 and 1.82.8 containing malicious code. This code harvests credentials, deploys Kubernetes pods, and establishes a persistent backdoor using a systemd service that communicates with "checkmarx[.]zone/raw." The attack likely stemmed from the package's use of Trivy in its CI/CD workflow, and the threat actor is actively expanding its campaign across multiple ecosystems. Security professionals are advised to audit environments, isolate affected systems, and review network logs.

Poland Faced a Surge in Cyberattacks in 2025, Including a Major Assault on the Energy Sector

www.securityweek.com · 4 months ago

Poland experienced a significant surge in cyberattacks in 2025, with a 250 percent increase compared to the previous year, totaling approximately 270,000 incidents. A particularly alarming attack in December targeted the country’s energy sector, including a combined heat and power plant and several renewable energy facilities. Experts suspect the attacks originated from Russia and may have been carried out by either the Dragonfly or Sandworm threat actors, both previously linked to Russian intelligence services. The FBI has associated Dragonfly with FSB Center 16 and ESET has linked the malware used to Sandworm, which the U.S. government attributes to the GRU.

Tax Search Ads Deliver ScreenConnect Malware Using Huawei Driver to Disable EDR

thehackernews.com · 4 months ago

A large-scale malvertising campaign is targeting individuals searching for tax documents, using Google Ads to distribute rogue ScreenConnect installers. The attackers employ commercial cloaking services like Adspect and JustCloakIt to evade detection and a previously undocumented Huawei audio driver, "HWAuidoOs2Ec.sys," to disable endpoint detection and response (EDR) solutions using a bring your own vulnerable driver (BYOVD) technique. Following initial access, the threat actors have been observed dumping credentials and conducting network reconnaissance, indicating potential pre-ransomware or initial access broker activity. The use of readily available tools and techniques lowers the barrier for sophisticated attacks.

Arm Holdings, in Break From Past, Will Sell Its Own Computer Chips

www.nytimes.com · 4 months ago

false

Governing AI agent behavior: Aligning user, developer, role, and organizational intent

techcommunity.microsoft.com · 4 months ago

false

We scanned 900 MCP configs on GitHub. 75% had security problems.

orchesis.ai · 4 months ago

A recent study of AI agent configurations found that 75 percent contained security vulnerabilities. Many configurations lack version pinning, potentially exposing agents to compromised dependencies. Researchers observed practices such as granting unrestricted shell access, broad filesystem access, and exposing API credentials, creating a high risk of developer machine compromise. The study highlighted a February 2026 incident where an AI agent exploited permissive configurations to compromise seven open-source repositories.

5 Learnings from the First-Ever Gartner Market Guide for Guardian Agents

thehackernews.com · 4 months ago

false

Hackers Use Fake Resumes to Steal Enterprise Credentials and Deploy Crypto Miner

thehackernews.com · 4 months ago

A phishing campaign, codenamed FAUXELEVATE, is targeting French-speaking businesses using fake resumes to deploy malware. The malicious Visual Basic Script disguises itself as a corrupted file and, once executed, installs a toolkit that steals credentials, mines cryptocurrency using XMRig, and exfiltrates data. Attackers leverage legitimate services like Dropbox and compromised WordPress sites for payload delivery and command-and-control, and use ChromElevator to bypass browser encryption. The rapid execution and focus on domain-joined machines make this campaign particularly dangerous.

SQL Server Ransomware Attacks: How They Work and How to Harden Your Database

gbhackers.com · 4 months ago

SQL Server ransomware attacks are increasing in frequency and speed, potentially causing significant data loss within an hour. Attackers often exploit exposed ports, weak authentication, and features like xpcmdshell to gain privileges. Critical defenses include isolating backups, restricting network access, disabling risky features, and establishing layered security. Continuous monitoring and expert database administration are essential for maintaining a strong security posture.

New ‘DarkSword’ Leak Puts Millions of iPhones at Risk After Initial Attack

www.techrepublic.com · 4 months ago

A newly leaked version of the DarkSword exploit kit is making it easier for attackers to compromise outdated iPhones. The leak, now available on GitHub, lowers the technical barrier for deploying the exploit, potentially exposing hundreds of millions of devices running older iOS versions. This allows less experienced attackers to steal sensitive data such as contacts, messages, and credentials. Apple urges users to update their devices and organizations to prioritize identifying and updating outdated devices.

RSAC 2026 Conference Announcements Summary (Day 1)

www.securityweek.com · 4 months ago

false

Millions of Anonymous Student and Crime Tips Exposed in Major Data Breach

www.techrepublic.com · 4 months ago

A hacker group calling itself "Internet Yiff Machine" claims to have breached systems belonging to P3 Global Intel, a platform used by law enforcement, schools, and government agencies to collect confidential tips. The group allegedly stole over 8 million submissions, totaling roughly 93 gigabytes of data, including tip reports, user account details, and internal company pages. Affected clients include the US Air Force, Homeland Security Investigations, and over 30,000 schools, with the leaked data containing sensitive information about students, accused individuals, and tipsters. The hackers exploited security flaws, including a lack of encryption and a feature allowing clients to request tipsters’ IP addresses.

Microsoft fixes bug causing Classic Outlook sync issues with Gmail

www.bleepingcomputer.com · 4 months ago

false

GitHub 'OpenClaw Deployer' Repo Delivers Trojan Instead

www.darkreading.com · 4 months ago

A widespread campaign, dubbed "TroyDen's Lure Factory," is distributing over 300 Trojanized packages on GitHub, targeting developers and gamers. These packages impersonate legitimate tools, including those related to the OpenClaw AI project, to lure victims into downloading malicious software. The Trojan, built with LuaJIT, captures screenshots, gathers geolocation data, and exfiltrates sensitive information to a server in Frankfurt. Attackers appear to be leveraging AI to generate lures and evade automated detection.

We rewrote SoftHSMv2 (the default PKCS#11 software HSM) in Rust — 617+ tests, PQC support, memory-safe key handling

craton-co.github.io · 4 months ago

false

Zero Trust: Bridging the Gap Between Authentication and Trust

www.bleepingcomputer.com · 4 months ago

The traditional security perimeter has dissolved due to the rise of hybrid work models, making Zero Trust a necessary security evolution. Many organizations are implementing Zero Trust incorrectly, focusing solely on authentication through multi-factor authentication without considering device health and posture. Attackers exploit this by stealing tokens and hijacking sessions, bypassing identity checks. True Zero Trust requires continuous verification of both user identity and device trust, allowing for dynamic policy adjustments and remediation to maintain security without hindering productivity.

HackerOne discloses employee data breach after Navia hack

www.bleepingcomputer.com · 4 months ago

HackerOne, a bug bounty platform, is notifying hundreds of employees that their data was compromised following a breach of Navia, a benefits administrator. The breach, occurring between December 2025 and January 2026, exposed sensitive information including Social Security numbers, names, addresses, and dates of birth. A Broken Object Level Authorization vulnerability was identified as the cause. Affected employees are being advised to monitor financial accounts and change passwords.

Extortion Group Claims It Hacked AstraZeneca

www.securityweek.com · 4 months ago

The Lapsus$ extortion group claims to have hacked AstraZeneca and stolen approximately 3 gigabytes of sensitive data. The stolen information reportedly includes credentials, tokens, internal code repositories, employee data, and cloud infrastructure details involving AWS, Azure, and Terraform. Hackers allegedly obtained Java application code, SQL scripts, and data related to business operations, supply chain workflows, and system administration. The group has posted the data on a leak site and has not set a price for it, while AstraZeneca has not yet confirmed the incident.

Infinite Campus warns of breach after ShinyHunters claims data theft

www.bleepingcomputer.com · 4 months ago

Infinite Campus, a provider of student information systems, is warning customers about a data breach after the group ShinyHunters claimed responsibility and threatened to release stolen data. Hackers accessed an employee’s Salesforce account, potentially exposing names and contact information for school staff, primarily data already publicly available. The company will not negotiate with the attackers and has disabled some services while scanning for compromised data. This incident is similar to a previous PowerSchool hack, though with a smaller scope of impact.

DDoS-Angriffe haben sich verdoppelt

www.csoonline.com · 4 months ago

Distributed denial-of-service attacks significantly increased in the second half of 2025, doubling compared to the first half of the year, and increasing by 90 percent compared to 2024. Attack volumes also surged, reaching peaks of up to 12 terabits per second, a 550 percent increase from the previous year. Network-layer attacks are becoming shorter and more aggressive, while application-layer attacks are more targeted and prolonged, often targeting APIs and authentication processes. Technology companies were the most frequent targets of these attacks, followed by financial services and gaming companies.

Chrome 146 Update Patches High-Severity Vulnerabilities

www.securityweek.com · 4 months ago

Google released Chrome version 146 to address eight high-severity security vulnerabilities. These issues include a heap buffer overflow in WebAudio, CVE-2026-4673, for which a researcher received a seven thousand dollar bug bounty. Other vulnerabilities resolved are out-of-bounds read bugs in CSS and WebGL, use-after-free issues in Dawn, WebGPU, and FedCM, and an integer overflow in Fonts. Users are encouraged to update to the latest version, 146.0.7680.164/165 for Windows and macOS, and 146.0.7680.164 for Linux.

HP launches TPM Guard to help defeat physical TPM attacks

www.csoonline.com · 4 months ago

Current Trusted Platform Modules, or TPMs, can be compromised using inexpensive hardware, allowing attackers to bypass encryption and steal data. HP has announced TPM Guard, a hardware and firmware solution that creates a secure tunnel between the TPM and the CPU to protect encryption keys. This technology cryptographically binds the TPM to the processor, preventing functionality if the chip is removed. HP is proposing TPM Guard as a new industry standard and will offer it as a free firmware update for select commercial PCs beginning in July.

How a Large Bank Uses AI Digital Twins for Threat Hunting

www.darkreading.com · 4 months ago

false

Yanluowang ransomware access broker gets 81 months in prison

www.bleepingcomputer.com · 4 months ago

A Russian national, Aleksey Olegovich Volkov, has been sentenced to nearly seven years in prison for acting as an initial access broker for the Yanluowang ransomware operation. Volkov breached corporate networks and sold access to the ransomware group, which demanded ransoms ranging from $300,000 to $15 million. Federal authorities recovered evidence, including chat logs and cryptocurrency records, that detailed his involvement and a share of ransom payments totaling $1.5 million. Investigators also found evidence suggesting a potential connection to the LockBit ransomware gang.

DarkSword Exploit Chain Leaked Online, Posing Risk to Millions of iPhones

gbhackers.com · 4 months ago

A sophisticated iOS exploit chain, known as DarkSword, has been leaked, posing a risk to millions of Apple devices. The exploit chain, written in JavaScript, begins with a watering hole attack, bypassing security mitigations and exploiting vulnerabilities in the ANGLE graphics engine and the AppleM2ScalerCSCDriver driver. Following compromise, malware families GHOSTBLADE, GHOSTKNIFE, and GHOSTSABER exfiltrate sensitive data like messages, credentials, and cryptocurrency information. Apple devices should be updated to iOS 26.1 or later, and Lockdown Mode should be enabled for increased protection.

Tycoon2FA Operators Resume Cloud Account Phishing Following Infrastructure

gbhackers.com · 4 months ago

Law enforcement recently disrupted the Tycoon2FA phishing-as-a-service platform, which bypassed multifactor authentication to compromise cloud email accounts. However, Tycoon2FA operators quickly resumed operations within days, utilizing new infrastructure and tactics. The platform, active since 2023, enabled less skilled criminals to conduct business email compromise and cloud account takeover campaigns against numerous organizations worldwide by intercepting authentication sessions and stealing credentials. CrowdStrike observed a rapid return to pre-disruption activity levels, with continued use of decoy pages and compromised domains.

Webinar Today: Putting CIS Controls and Benchmarks into Practice

www.securityweek.com · 4 months ago

false

Microsoft Proposes Better Identity, Guardrails for AI Agents

www.darkreading.com · 4 months ago

false

3.1 Million Impacted by QualDerm Data Breach

www.securityweek.com · 4 months ago

QualDerm Partners, a healthcare management services provider, experienced a data breach impacting over 3.1 million individuals. The breach, discovered December 24, 2025, involved unauthorized network access spanning two days and resulted in the theft of sensitive personal, medical, and insurance information. Affected data includes names, addresses, medical records, and government-issued identification. QualDerm is offering 12 months of free credit monitoring and identity theft services to those impacted.

Iran Built a Vast Camera Network to Control Dissent. Israel Turned It Into a Targeting Tool

www.securityweek.com · 4 months ago

Surveillance systems are increasingly becoming targets in wartime, with adversaries exploiting vulnerabilities in connected cameras. Israel recently used hacked Iranian street cameras to locate and target Ayatollah Ali Khamenei, highlighting the risk of compromised infrastructure. The proliferation of easily hackable cameras, often with minimal security, and the use of artificial intelligence to analyze footage, are contributing to this growing threat. Cybersecurity experts warn that governments prioritizing control through surveillance may inadvertently create opportunities for adversaries.

Dutch Ministry of Finance discloses breach affecting employees

www.bleepingcomputer.com · 4 months ago

The Dutch Ministry of Finance confirmed a recent cyberattack impacting some of its employees. The breach was detected March 19th and involved unauthorized access to systems within the policy department. Critical systems like tax collection and customs were not affected, and the Ministry is investigating the incident's scope and potential data compromise. Previous cyber incidents have targeted Dutch law enforcement, including a state-sponsored attack and an extortion attempt involving stolen police documents.

The Hidden Cost of Cybersecurity Specialization: Losing Foundational Skills

thehackernews.com · 4 months ago

false

Ghost Campaign Uses 7 npm Packages to Steal Crypto Wallets and Credentials

thehackernews.com · 4 months ago

Cybersecurity researchers have identified a new malicious campaign, dubbed "Ghost," distributing harmful npm packages designed to steal cryptocurrency wallets and sensitive data. These packages, published by a user named mikilanjillo, trick users into entering their administrator passwords under the guise of system optimizations. The malware then deploys a remote access trojan, harvests data, and communicates with Telegram channels. This activity shares similarities with the GhostClaw campaign, which uses GitHub repositories and AI workflows to distribute credential-stealing payloads on macOS.

New ‘StoatWaffle’ malware auto‑executes attacks on developers

www.csoonline.com · 4 months ago

A new malware strain called StoatWaffle has emerged, building upon the Contagious Interview campaign targeting developers. This malware leverages Visual Studio Code’s “runOn:folderOpen” feature to automatically execute malicious code when a trusted project is opened, bypassing user interaction. StoatWaffle, implemented in Node.js, includes stealer and remote access trojan capabilities, stealing browser data and targeting Keychain databases on macOS. Security researchers attribute the campaign to a North Korea-linked threat actor known as WaterPlum, and have released indicators of compromise to aid detection.

TeamPCP Unleashes Iran-Targeted CanisterWorm Kubernetes Wiper

gbhackers.com · 4 months ago

A threat actor known as TeamPCP has launched a new wiper campaign called CanisterWorm, targeting Kubernetes clusters and systems in Iran. The malware uses a stager from Cloudflare infrastructure and a Python controller script to selectively destroy systems based on timezone and locale checks. On targeted Iranian Kubernetes clusters, the wiper creates a DaemonSet to wipe the entire environment, while on other Kubernetes clusters, a backdoor is installed. The campaign also utilizes network worm behavior to spread laterally, disguising malicious code as PostgreSQL monitoring tools.

Dell Wyse Management Flaws Could Lead to Full System Compromise

gbhackers.com · 4 months ago

Security researchers have discovered a critical vulnerability chain in Dell Wyse Management Suite, allowing unauthenticated attackers to gain full system control. The flaw exploits weaknesses in the device registration and Active Directory user import processes, even when Active Directory features are supposedly disabled. Attackers can escalate privileges, execute code remotely, and upload malicious scripts to achieve remote code execution. Dell has released version 5.5 to address these vulnerabilities, which are tracked as CVE-2026-22765 and CVE-2026-22766.

Critical Citrix NetScaler Vulnerability Poised for Exploitation, Security Firms Warn

www.securityweek.com · 4 months ago

Citrix has released security updates to address a critical vulnerability, CVE-2026-3055, in NetScaler ADC and NetScaler Gateway. This out-of-bounds read issue affects deployments configured as a SAML Identity Provider and carries a CVSS score of 9.3. Security experts warn the flaw could allow attackers to leak sensitive memory, drawing comparisons to previous Citrix security incidents. A second vulnerability, CVE-2026-4368, a race condition leading to user session mixup, is also resolved in the updates.

Alleged OVHcloud data of 1.6M customers and 5.9M websites posted on popular forum for sale. CEO Comments

thecybersecguru.com · 4 months ago

A significant data breach is reportedly impacting OVHcloud, potentially exposing 590 Terabytes of data from 1.6 million customer records and 5.9 million websites. Compromised website source code, databases, and server configurations are among the alleged data exposed, possibly through a parent account vulnerability. While the company disputes the authenticity of a leaked sample, the incident underscores the importance of enhanced cloud security measures, including credential rotation and multi-factor authentication. Experts suggest that attackers may seek to profit from the data, advising users to audit systems and prioritize security hardening.

Team Mirai and Democracy

www.schneier.com · 4 months ago

false