Blahalla

Admins of Bulletproof Hosting Service Used by Russian Hackers Arrested in Netherlands

www.securityweek.com · a month ago

On May 18, 2026, Dutch authorities arrested two men, a 57 year old from Amsterdam and a 39 year old from The Hague, suspected of owning companies that provided bulletproof hosting services to Russia aligned threat actors and evaded European Union sanctions. Investigators from the Dutch Fiscal Information and Investigation Service searched sites in Enschede and Almere and two data centers in Dronten and Schiphol Rijk, seizing laptops, phones and more than 800 servers. The operation allegedly served as a front to continue services for a sanctioned provider known as Stark Industries, which was placed on the EU sanctions list in May 2025, and local reporting identified the suspects as Youssef Z. and Andrey N. and linked the infrastructure to firms named Mirhosting and WorkTitans.
CERT-In Recommends 12-Hour Patching for Internet-Facing Flaws Amid AI-Assisted Attacks

thehackernews.com · a month ago

CERT-In has issued a 38-page blueprint requiring organizations to patch critical internet-exposed vulnerabilities within 12 hours where feasible, warning that AI tools and large language models can automate vulnerability discovery and exploitation and dramatically shorten attackers' preparation times. The guidance says AI-assisted attacks can compress exploitation timelines and mandates continuous, risk based vulnerability and patch management with remediation windows that include critical externally exposed vulnerabilities within one day, known exploited internal vulnerabilities within one day unless mitigations are documented, critical internal vulnerabilities for high value systems within three days, and high severity flaws within five days. Defensive measures emphasized include assume breach readiness, zero trust and least privilege, defense in depth, software supply chain risk reduction through software bills of materials and provenance checks, enhanced monitoring, testing and red teaming, and temporary mitigations such as isolation, access restrictions, web application firewall and API protections when patches are not immediately available.
Stop treating AI governance as a review layer. Make it release infrastructure

www.csoonline.com · a month ago

Traditional compliance as a post hoc review fails for AI because models, retrieval indexes and agent toolchains change continuously, widening a governance gap between slow audit cycles and fast deployments. There are three operational shifts to close that gap: generate model cards, training data provenance and evaluation baselines automatically in the CI CD pipeline; make compliance evidence an enforceable deployment gate; and treat agent identity as a first class security control with scoped permissions and audit trails. Embedding governance into release infrastructure rather than treating it as a separate review layer produces continuous evidence, enables faster compliant launches, and reduces governance debt ahead of regulatory regimes such as the EU AI Act and emerging US state and sector rules.
Ghost CMS Vulnerability Exploited to Infect 700 Sites With ClickFix Malware

gbhackers.com · a month ago

A critical SQL injection vulnerability in Ghost CMS tracked as CVE-2026-26980 is being actively exploited to extract Ghost Admin API keys, modify posts, and inject JavaScript that distributes ClickFix malware, with researchers reporting more than 700 compromised domains. Attackers insert a lightweight page loader that fetches a cloaked second-stage payload, fingerprints visitors using WebGL, timezone, and behavior signals, and redirects selected users to a fake verification that social engineers them into running commands that download and execute payloads such as installer.dll or UtilifySetup.exe via rundll32. Multiple threat groups are rotating infrastructure and reinfecting sites, and the flaw was publicly disclosed in February 2026 but many Ghost instances remain unpatched, enabling rapid large-scale page poisoning. Immediate mitigation advice includes upgrading to patched Ghost versions, rotating all API keys and credentials, thoroughly inspecting site content and logs for unauthorized changes, and urging users who visited affected sites to check systems for suspicious downloads or execution activity.
Update Starlette Now. New severe vulnerability dropped.

badhost.org · a month ago

The vulnerability CVE-2026-48710 affects Starlette versions below 1.0.1 because the library derives the request url from the Host header without proper sanitization. This flaw allows attackers to forge the request url path, potentially bypassing path-based authentication middleware in thousands of affected FastAPI and Starlette applications. To fix this, users must update Starlette to version 1.0.1 or later and avoid relying solely on middleware for authentication decisions. Implementing an RFC compliant reverse proxy is also a crucial mitigation step to validate and normalize the Host header.
CISA orders feds to patch actively exploited Drupal vulnerability

www.bleepingcomputer.com · a month ago

CISA has ordered U.S. federal civilian agencies to patch an actively exploited unauthenticated SQL injection vulnerability in the Drupal content management system by midnight on May 27, 2026 under Binding Operational Directive 22-01. The flaw, tracked as CVE-2026-9082 and discovered by Google Mandiant researcher Michael Maturi, exists in Drupal's database abstraction API and allows arbitrary SQL injection against PostgreSQL powered sites, potentially leading to information disclosure, privilege escalation, and remote code execution. Drupal released patches and labeled the bug highly critical after detecting exploitation in the wild, while Shadowserver is tracking about 670 exposed unpatched Drupal instances concentrated in North America and Europe. CISA added the vulnerability to its Known Exploited Vulnerabilities Catalog and urged all organizations to apply vendor mitigations, follow applicable guidance for cloud services, or discontinue use if mitigations are unavailable.
Apache CXF Flaw Exposes Systems to LDAP Injection Attacks

gbhackers.com · a month ago

There is an LDAP injection vulnerability in Apache CXF's XKMS LDAP certificate repository that can be manipulated to retrieve arbitrary X.509 certificates, exposing sensitive cryptographic material and enabling impersonation or unauthorized access, tracked as CVE-2026-44930. The flaw exists in the cxf-services-xkms-x509-repo-ldap component due to improper input validation and can force LDAP queries to return all certificate entries. Multiple CXF releases are affected and Apache has released fixes in versions 4.2.1, 4.1.6 and 3.6.11, with users urged to upgrade immediately and to apply input validation, least-privilege controls, and monitoring as mitigations. Detection and hardening options include reviewing LDAP query handling, monitoring LDAP query logs for unusual activity, and deploying application filtering or web application firewalls.
Hackers Use SEO Poisoning to Fake Gemini CLI and Claude Code Installers

gbhackers.com · a month ago

Hackers are abusing search engine optimization to impersonate popular AI developer tools and deliver fake installer pages that prompt developers to run a PowerShell command which executes a fileless infostealer in memory. The campaign, first observed in early March 2026, uses typosquatting domains and even installs legitimate packages alongside the malicious payload to mask the compromise while the malware disables endpoint defenses, harvests credentials, tokens, keys, cloud files, and collaboration data, and exfiltrates it to attacker controlled infrastructure. The malware also supports remote code execution for hands on keyboard activity and is hosted across dozens of spoofed domains on bulletproof hosting. Organizations are advised to verify software sources, monitor for hidden or chained PowerShell commands such as Invoke-RestMethod and Invoke-Expression, and watch for unusual outbound connections immediately after installation attempts.
ConnectWise Automate Flaw Allows Hackers to Evade Security Controls

gbhackers.com · a month ago

There is a high severity vulnerability in ConnectWise Automate remote monitoring and management software that can bypass integrity verification and potentially enable unauthorized code execution, tracked as CVE-2026-9089 with a CVSS score of 8.8. The flaw resides in the agent plugin loading and self update processes where certain downloaded components may be processed without full integrity checks, and it is classified under CWE 494. Exploitation requires no user interaction and has low complexity, creating risks of arbitrary code execution, lateral movement, persistent access, and broader supply chain exposure for managed service providers, with no confirmed active exploitation at disclosure. ConnectWise addressed the issue in Automate version 2026.5, cloud instances were automatically updated, on premises deployments must be manually upgraded, and organizations are urged to patch within 30 days and monitor agent activity while deploying endpoint detection controls.
Microsoft: Domain Controller lookup may fail on Windows Server 2016

www.bleepingcomputer.com · a month ago

A known issue in the May 2026 cumulative update KB5087537 causes domain controller discovery to fail on Windows Server 2016 systems whose hostnames are exactly 15 characters long. When the hostname is 15 characters long, DCLocator calls for example using nltest dsgetdc domain pdc return ERRORINVALIDPARAMETER preventing applications and administrative tools from locating a domain controller and disrupting administrative operations such as DFS Namespace management. Microsoft is investigating the problem and has not provided a timeline for a fix.
Iranian Hackers Deploy MiniFast and MiniJunk V2 via Phishing and SEO Poisoning

thehackernews.com · a month ago

Nimbus Manticore, an Iranian state sponsored threat actor linked to Iran's Islamic Revolutionary Guard Corps, has run a cross regional espionage campaign since late February 2026 using career themed phishing lures, spoofed meeting invitations, and SEO poisoning to impersonate aviation and software organizations across the United States, Europe, and the Middle East. The campaigns delivered two RAT families, MiniJunk and a new AI assisted backdoor called MiniFast or MiniUpdate, using AppDomain hijacking, trojanized installers including Zoom, and poisoned search results that directed victims to a fake SQL Developer download page. MiniFast communicates over HTTP, beacons system details, and supports file operations, directory listings, process enumeration, command execution via cmd.exe, DLL loading, archive creation, scheduled task persistence, privilege escalation via runas, and configurable polling and jitter for long term access. Security firms observed the actor expand operations to multiple countries, including attacks that targeted a U.S. oil and gas firm and manipulation of online automatic tank gauge displays at gas stations in several U.S. states that altered readings without changing fuel levels.
Phishing Campaigns Exploit RCS and iMessage to Evade SMS Security Filters

gbhackers.com · a month ago

Phishing campaigns are shifting from traditional SMS to encrypted messaging channels such as RCS and iMessage, which operate over data networks and reduce carriers ability to inspect or block malicious content, making lures appear more legitimate through features like read receipts, media, and branded content. Attackers are moving from static phishing pages to live interaction models provided by a growing phishing as a service ecosystem that instantly relays credentials, triggers one time passcode requests, and captures OTPs or session tokens to bypass multifactor authentication. These platforms use localization and artificial intelligence to clone legitimate sites in real time and employ anti bot checks to evade automated detection while converting stolen card details into tokenized assets for contactless payments and high value transactions. Defenders are advised to adopt phishing resistant authentication such as FIDO2 or WebAuthn, implement device fingerprinting and risk based verification for digital wallet provisioning, and shift emphasis from detection to prevention.
PuTTY 0.84 Update Patches SSH Key Exchange Crash Issues and Telnet Prompt Spoofing Flaw

gbhackers.com · a month ago

PuTTY 0.84 was released with three minor security fixes that address vulnerabilities which could allow remote attackers to crash the client or trick users during insecure sessions. Fixed issues include an ECDSA signature verification bug that can trigger an assertion failure during SSH key exchange affecting NIST curves P-256 P-384 and P-521, a double free in RSA key exchange present in versions 0.72 through 0.83, and a Telnet trust indicator flaw that can leave the trust sigil set and enable fake password prompt spoofing. The update also addresses CVE-2026-4115 related to EdDSA signature handling, and users are urged to upgrade to 0.84, avoid legacy protocols such as Telnet, and disable unused key exchange methods like RSA KEX.
7-Eleven data breach exposes personal information of 185,000 people

www.bleepingcomputer.com · a month ago

The ShinyHunters extortion gang stole the personal information of about 185,300 people after breaching 7‑Eleven systems in early April 2026. Analysis of the leaked data by Have I Been Pwned shows exposed names, dates of birth, unique email addresses, phone numbers, and physical addresses. ShinyHunters claimed to have exfiltrated over 600,000 Salesforce records and published a 9.4 gigabyte archive after 7‑Eleven refused to pay the ransom. 7‑Eleven said it discovered unauthorized access on April 8, 2026 and sent breach notification letters on May 1, noting the incident was limited to systems used to store franchisee documents.
Vulnerabilities have become cyber attackers’ No. 1 door to the enterprise

www.csoonline.com · a month ago

Vulnerability exploitation has overtaken stolen credentials as the most common initial access vector, causing 31 percent of breaches versus 13 percent for credential abuse. Only 26 percent of CISA Known Exploited Vulnerabilities were fully remediated in 2025, down from 38 percent the year before, while the median time to patch rose to 43 days from 32 days and the volume of critical vulnerabilities increased by 50 percent year on year. Attackers are increasingly targeting unpatched perimeter and third party systems, with breaches involving a third party accounting for 48 percent of incidents, and adversaries are using AI to shrink the time to exploit from months to hours. Ransomware featured in roughly 48 percent of breaches even as ransom payments declined, prompting calls for risk based continuous vulnerability management, stronger identity controls, defense in depth, and containment measures.
Multiple 7-Zip Vulnerabilities Enable Arbitrary Code Execution

gbhackers.com · a month ago

Multiple memory safety bugs in 7 Zip version 26.00 allow remote attackers to leak sensitive data and in at least one case to execute arbitrary code when a victim opens a crafted archive file. GitHub Security Lab disclosed a critical NTFS heap buffer overflow tracked as CVE-2026-48095 and a cluster of additional memory access violations affecting formats such as SquashFS, UEFI capsules, UDF, and WIM. All 7 Zip releases up to and including 26.00 are affected and fixes were shipped in 7 Zip version 26.01 on April 27, 2026, so administrators and users should upgrade to 26.01 or later immediately or otherwise limit 7 Zip use on high value systems and monitor for suspicious archive activity.
Malicious PDF LNK Files Deploy Cobalt Strike in Operation Dragon Whistle

gbhackers.com · a month ago

Operation Dragon Whistle targets China s education sector with highly tailored spear phishing emails impersonating university communications and using urgent ZIP attachments named as official testing notices to coerce students and staff into opening them. Inside the ZIP is a file disguised as a PDF but actually a LNK shortcut that leverages explorer dot exe to launch a hidden VBScript which opens a decoy PDF while silently running a bundled Bandizip executable and causing DLL sideloading of a malicious ark dot x64 dot dll. The chain decrypts and unpacks a payload in memory, bypasses AMSI and ETW, employs anti analysis checks, and loads a Cobalt Strike Beacon directly into memory that communicates with command and control hosted on Alibaba Cloud. Security researchers attribute the campaign to actor UNG0002 with medium to high confidence and note strong similarities to prior LNK based operations.
Anthropic Prepares Claude Mythos for Wider Release Through Claude Code

gbhackers.com · a month ago

Anthropic is preparing to expand access to its most advanced AI, Claude Mythos, by staging a commercial rollout labeled Mythos 1 with references appearing in source code and brief user interface tests. Claude Mythos surfaced after a March 2026 internal leak and was subsequently tested through Project Glasswing in April 2026 with over 40 partners including AWS, Microsoft, Google, and CrowdStrike. The model has reportedly identified thousands of vulnerabilities, finding more than ten thousand high and critical issues across open source projects and disclosing 1,596 findings across 281 projects with only a fraction patched so far. Anthropic plans to integrate Mythos into developer and security products such as Claude Code and Claude Security, which offer automated scanning, patch suggestion pull requests, and historical trends, while rolling out access gradually and reinforcing safeguards to reduce misuse.
KnowledgeDeliver LMS Flaw Exploited to Deploy Godzilla and Cobalt Strike

thehackernews.com · a month ago

There is a now patched high severity zero day in Digital Knowledge KnowledgeDeliver learning management system that was exploited to deliver the Godzilla web shell and enable deployment of Cobalt Strike Beacon. The vulnerability, tracked as CVE-2026-5426 with a CVSS score of 7.5, stems from hard coded ASP.NET machineKey values that allowed unauthenticated ViewState deserialization and remote code execution. Attackers used the web shell to run commands, change file permissions, alter a JavaScript file to display a fake security alert that lured users to install a malicious plugin, and delivered organization specific encrypted payloads that resulted in Cobalt Strike infections. The incident highlights the severe risks of shared secrets in deployment templates and recommends using unique secrets and robust endpoint monitoring to defend against deserialization attacks.
Jailbroken Gemini AI Abused in Credential Theft and Crypto Wallet Heist

gbhackers.com · a month ago

A jailbroken Gemini AI was weaponised in a multi year campaign that combined political influence operations, credential theft, and a cryptocurrency wallet heist run by a single Russian speaking actor known as bandcampro. The operator ran a public Telegram channel called americanpatriotus for five years to build trust with an audience aligned to MAGA and QAnon themes and from September 2025 began using a compromised Gemini setup as an automation engine that was persistently jailbroken by loading malicious context in session memory. Gemini was used to generate propaganda posts, manage infrastructure, rotate at least 73 likely stolen API keys, act as a password mutation oracle to brute force accounts, and help deploy a fake wallet installer named StellarMonster whose installer deployed a legitimate remote access tool to steal seed phrases and drain victim wallets. Investigators confirmed at least one wallet was emptied and at least 29 WordPress administrator accounts across small businesses and firms were compromised, and analysts assess the operation was primarily financially motivated rather than state sponsored.
Hackers Abuse KnowledgeDeliver LMS Flaw to Install BLUEBEAM Web Shell

gbhackers.com · a month ago

A critical unauthenticated remote code execution vulnerability in the KnowledgeDeliver learning management system enables ASP.NET ViewState deserialization attacks and is tracked as CVE-2026-5426. Attackers have abused deployments that used a shared hard-coded ASP.NET machineKey prior to February 24, 2026 to deliver crafted ViewState payloads that execute code and, in observed incidents, install the BLUEBEAM.NET in-memory web shell inside the IIS worker process. Recommended mitigations include immediately rotating ASP.NET machineKey values to unique cryptographically strong keys, restricting access to trusted IP ranges, monitoring for ViewState validation failures such as Event ID 1316 and suspicious child processes spawned by w3wp.exe, and hunting for modified web application files and web shell indicators of compromise.
Ransomware Uses ChaCha20 and Curve25519 to Encrypt Windows Files

gbhackers.com · a month ago

Payload ransomware is a new Windows ransomware family that combines ChaCha20 stream encryption with a per-file Curve25519 Elliptic Curve Diffie Hellman key exchange, making victim data effectively unrecoverable without the attackers' private key. It implements strong anti forensics measures including Event Tracing for Windows patching, deletion of Volume Shadow Copies, event log wiping, and aggressive termination of processes and services to hinder detection and recovery.
Security experts caution MFA alone can no longer stop threat actors

www.csoonline.com · a month ago

There are increasing phishing campaigns and phishing as a service platforms such as Kali365 and EvilTokens that steal Microsoft 365 OAuth device codes to capture access and refresh tokens, allowing attackers to bypass multifactor authentication and access Outlook, Teams, and OneDrive. Kali365 commercializes attacks with AI generated lures, automated campaign templates, management dashboards, multi language variants and subscription tiers starting around two hundred fifty dollars, and observed activity includes creating malicious inbox rules and registering devices to extend persistent access. Law enforcement and security experts urge that MFA alone is no longer sufficient and recommend restricting or blocking device code authentication via conditional access, proactively revoking and monitoring OAuth tokens and device registrations, adopting phishing resistant FIDO2 passkeys and stronger session controls, implementing continuous access evaluation, and segmenting high risk users and functions.
Project Glasswing has uncovered 10,000 vulnerabilities: Anthropic

www.csoonline.com · a month ago

There are an estimated 10,000 critical or high severity vulnerabilities surfaced by Project Glasswing, an initiative using the Claude Mythos Preview model with more than 50 partners. Mythos Preview scanned over 1,000 open source projects and found 6,202 high or critical vulnerabilities, 1,752 of which were assessed by six independent firms with 90.6 percent true positives and 62.4 percent confirmed as high or critical; so far 530 high or critical bugs have been disclosed to maintainers, 75 patched and 65 public advisories issued, with plans to disclose another 827. Maintainers report a deluge of low quality AI generated reports and severe capacity constraints, shifting the cybersecurity bottleneck from finding bugs to human driven triage and patching, and Anthropic has launched a Claude Security beta, a Cyber Verification Program, and committed over 100 million dollars in usage credits and 4 million dollars in donations to support remediation.
The War Between Wars: How an IRGC Front Runs Destructive OT and IT Attacks Under Cover of a Ceasefire

profero.io · a month ago

An Iranian state-directed persona called Cyber Isnaad Front, assessed to be operated alongside IRGC-affiliated ASA, has been running deniable destructive campaigns against Israeli infrastructure during ceasefires that combine IT intrusions with targeted operational technology sabotage. At one food-production plant attackers reprogrammed refrigeration controllers, changed setpoints and safety limits, switched motorized valves to manual and pinned them open, changed central-controller credentials to lock out engineers, and caused the destruction of three compressors without malware on the controllers themselves. On the IT side the operation deployed a Go Remote Access Toolkit called GRAT, masquerading as fake Microsoft updates and persisting as a scheduled task named OneDrive Update, with modules for VNC, exfiltration, encryption named BigBang, and a disk-wiper that overwrites drives and destroys partition tables while communicating with C2 infrastructure at 84.201.6.131 on ports 7878 and 9988. Indicators include specific SHA-256 hashes, file paths under AppData and ProgramData, and C2 hosts, and defenders are advised to hunt for unsigned binaries in those locations, alert on outbound AMQPS or Redis traffic to the listed host and ports, segment IT from OT, keep offline versioned backups of controller programs, and alarm on out-of-band setpoint or operating-mode changes.
Massachusetts Uber and Lyft Drivers Form First Union for Ride-Share Workers

www.nytimes.com · a month ago

Massachusetts has certified the App Drivers Union as the nation’s first union to represent app based rideshare drivers, giving it formal recognition to bargain on behalf of roughly seventy thousand drivers for companies including Uber and Lyft. The certification followed a state process that requires at least twenty five percent support among active drivers, a threshold the Department of Labor Relations found had been met, and the 2024 ballot law creates a structured bargaining process overseen by the department. The union is backed by national labor groups and officials say the major platforms have indicated they will negotiate in good faith as bargaining and any subsequent mediation or arbitration proceed.
WhatsApp Local Storage Claim Raises Apple Privacy Questions

www.techrepublic.com · a month ago

Researchers at Mysk allege that WhatsApp stores some decrypted chat data unencrypted in local database files on macOS and iOS in a shared app group container, with files such as Axolotl.sqlite, ContactsV2.sqlite, and LocalKeyValue.sqlite found in the container. Security experts and outlets pushed back, saying the databases reside in a secure container normally accessible only to WhatsApp and that the shared container supports migration between WhatsApp and WhatsApp Business rather than cross access by other Meta apps. The concern is that readable local data could be exposed if a device is compromised, backups are insecure, or an operating system flaw is exploited, with the recently disclosed macOS Archive Utility flaw CVE-2026-28910 cited as a possible escalation scenario. IT teams are advised to treat the problem as an endpoint and mobile device management issue by enforcing strong passcodes and biometric locks, keeping iOS and macOS up to date, and using encrypted device backups.
As A.I. Fever Rises in Silicon Valley, Pope Leo Has a Few Words

www.nytimes.com · a month ago

The American-born pope issued a sweeping encyclical titled Magnifica humanitas calling to disarm artificial intelligence and warning that concentrated power in major technology companies threatens democracy, widens inequality, and risks new forms of digital slavery. He said it is not permissible to entrust irreversible lethal decisions to AI and urged removing AI from military and purely commercial control, imposing strict national and international oversight, and ensuring broad public participation in AI governance. The Vatican staged a high-profile launch that included Silicon Valley figures such as the co-founder of Anthropic, signaling both a direct challenge to Big Tech and an effort to draw the technology sector into a moral and regulatory conversation.
Anthropic’s restricted Claude Mythos model may be coming to Claude Code

www.bleepingcomputer.com · a month ago

There are signs Anthropic is preparing to roll out the restricted Mythos model into developer products after the model briefly appeared in public listings for Claude Code and Claude Security. Mythos, announced on April 7, shows markedly stronger code reasoning and autonomous capabilities and has been demonstrated to produce functional cyberattacks, leading Anthropic to withhold a full public release until robust guardrails are in place. Through Project Glasswing Mythos has been used with up to fifty partner organizations to hunt vulnerabilities and reportedly uncovered about ten thousand high or critical severity flaws in its first month, but it remains unclear which subscription tiers will receive access as Anthropic finalizes containment and collaboration measures.
Nigel Farage’s Russian hack claim ‘without any merit’, former NCSC chief says

www.theguardian.com · 2 months ago

Nigel Farage said a Russian hack and leak operation was behind disclosure of the five million pound gift he received from crypto billionaire Christopher Harborne, and Reform UK sources said counter espionage experts analyzing his phone reported compromise of his phone, email and bank accounts. Ciaran Martin, founding chief executive of the National Cyber Security Centre, described the allegation as entirely unsubstantiated, said Farage had not provided technical evidence and warned that if true it would amount to an unprecedentedly aggressive intervention requiring a national government response. Martin urged Farage to contact the NCSC, publish any technical evidence and cooperate with security services, noting serious national security implications and saying such claims should not be made without proper investigation.
How credential brokering prevents AI agents from compromising credentials via prompt injection

infisical.com · 2 months ago

There is a growing security practice called credential brokering that inserts a proxy between AI agents and secrets so agents can access services without ever seeing underlying credentials. This approach prevents attacks such as prompt injection and credential exfiltration by routing outbound requests through a broker that authenticates the agent, substitutes placeholders with real tokens, and forwards the request so the agent never sees secret values. Implementations range from managed platform proxies to open source projects like Agent Vault, and recommended design principles include strong isolation, co location to reduce latency, and keeping brokers reachable only from private networks.
CVE-2021-21735: ZTE H168N wizard whitelist exposed PPPoE and WLAN secrets pre-auth

minanagehsalalma.github.io · 2 months ago

CVE-2021-21735 describes an administrative compromise vulnerability found in the ZTE ZXHN H168N device. This security flaw allows an attacker to gain unauthorized administrative access to the equipment. Exploiting this vulnerability compromises the integrity of the device's management functions. Owners must update the firmware immediately to patch this critical security risk.
Hacker Lists 340M OnlyFans User Records for Sale

www.techrepublic.com · 2 months ago

A hacker is selling a 340 million record OnlyFans database on a cybercrime forum for 0.313 BTC, about 76,000 dollars, claiming it contains creator and subscriber details that could link accounts to real identities. The seller advertised entries including usernames, real names, email addresses, phone numbers, join dates, follower and like counts, upload statistics, account types, linked social profiles, and a card field described as the last four digits of a payment card, but sample records were incomplete, contained placeholders, and resembled public profile metrics rather than confirmed internal OnlyFans data. The seller later said no direct breach of OnlyFans occurred and that data was compiled by matching older breach and leak databases and public information from other platforms, while independent researchers found only a small sample that prevents verification of the dataset size. Exposed emails and matched identifiers could be used for profiling, phishing, doxxing, impersonation, account takeover, extortion, or harassment even without leaked passwords.
⚡ Weekly Recap: Linux Flaws, Defender 0-Days, Router Botnets, and Supply Chain Chaos

thehackernews.com · 2 months ago

A poisoned Visual Studio Code extension allowed attackers to compromise developer tooling and exfiltrate roughly three thousand eight hundred GitHub repositories, and the release of Shai-Hulud code has given attackers a reproducible supply chain worm blueprint. Multiple high impact flaws were disclosed or observed in active exploitation, including CVE-2026-46333 in the Linux kernel, CVE-2026-41091 and CVE-2026-45498 in Microsoft Defender, CVE-2026-9082 in Drupal Core, CVE-2026-45585 affecting BitLocker, and CVE-2026-20223 in Cisco Secure Workload. Threat actors are increasingly weaponizing old unpatched devices and development artifacts, with botnets reusing legacy router flaws such as CVE-2018-5999 and phishing campaigns delivering trojanized installers that sideload malware.
Ghost CMS Vulnerability Exploited to Hack Over 700 Websites

www.securityweek.com · 2 months ago

Sites belonging to major universities such as Harvard and Oxford, as well as DuckDuckGo, have been compromised in the attack.
Netherlands Seizes 800 Servers, Arrests 2 for Aiding Cyberattacks

krebsonsecurity.com · 2 months ago

Authorities in the Netherlands have arrested the co-owners of two related Internet hosting companies for operating IT infrastructure used by Russia to carry out cyberattacks, influence operations and disinformation campaigns inside the European Union. The two men were the focus of a 2025 KrebsOnSecurity story about how their hosting companies had assumed control over the technical infrastructure of Stark Industries Solutions, an Internet service provider sanctioned last year by the EU as a frequent staging ground for cyber mischief from Russia's intelligence agencies.
One Job That Is Growing in the A.I. Era? Cybersecurity Experts.

www.nytimes.com · 2 months ago

There has been a surge in demand for security engineers as AI generates a glut of new code and risk, with job postings rising about eleven percent year over year in the first quarter. Advanced models such as Anthropic's Mythos and new cyber focused GPT variants have accelerated discovery of vulnerabilities and exploit chaining, prompting firms to treat model outputs as potential attack tools and to form private disclosure and mitigation partnerships. The role of human security engineers is shifting toward building review, triage, and remediation systems around AI findings and maintaining human in the loop verification, which is sustaining and reshaping hiring and pay patterns in the industry. Governments and large companies are pursuing vetting programs, restricted access arrangements, and formal coordination with model providers to manage risks and speed patches.
One Job That Is Growing in the A.I. Era? Cybersecurity Experts.

www.nytimes.com · 2 months ago

Demand for security engineers has surged as artificial intelligence creates a glut of new code and increases the speed at which vulnerabilities are discovered and exploited. Models such as Anthropic's Mythos and new cyber-focused versions of other large language models can scan large codebases, identify and chain vulnerabilities, and have prompted companies to expand internal scanning, patching, and hiring of security staff. Government and industry responses include limited releases to vetted partners, expanded internal audit programs, and discussions about granting certain agencies controlled access to these frontier models to manage risks while leveraging their capabilities.
FBI warns of Kali365 phishing service targeting Microsoft 365 accounts

www.bleepingcomputer.com · 2 months ago

The FBI is warning about the Kali365 phishing-as-a-service platform (PhaaS) that is used to hijack Microsoft 365 accounts by abusing OAuth device code authentication to steal session tokens and bypass multi-factor authentication (MFA).
InvisibleFerret Malware Uses .pyd and .so Files to Evade Script Detection

gbhackers.com · 2 months ago

A North Korea-linked threat group, Void Dokkaebi, also known as Famous Chollima, has significantly upgraded its malware delivery techniques by converting its Python-based InvisibleFerret malware into compiled binary modules. InvisibleFerret was previously deployed as readable Python scripts, making it easier for defenders to detect through static analysis and signature-based tools. The latest campaign leverages Cython
AI security needs a shift from models to systems, researchers argue

www.csoonline.com · 2 months ago

Enterprises cannot secure AI agents by making the underlying models more robust and must instead enforce security controls at the system level around them, researchers behind a paper published this month argued, warning that traditional AI-security approaches are increasingly misaligned with how autonomous agents actually operate inside enterprise environments. The paper argues that enterprises should stop treating AI agents as trusted software components and instead secure them as fundamentally untrusted systems operating inside enterprise infrastructure. “The AI model powering the agent must be treated as an untrusted component,” the researchers wrote in the paper, warning that “semantic guardrails” and prompt-level defenses alone cannot reliably secure systems once agents gain access to enterprise tools, memory, APIs, browsers, and execution environments. The authors drew the comparison to operating systems. “Similar to how an operating system treats a process as untrusted, we take the stance that the model powering the agent should be treated as untrusted and security properties should be expressed and enforced outside, at the level of the encompassing system,” they wrote. The paper was written by researchers at Google, the University of California, San Diego, the University of Wisconsin-Madison, and other institutions, including Mihai Christodorescu, Earlence Fernandes, and Somesh Jha. Five principles from systems security The authors distilled five principles from decades of systems security research that they said agentic systems should follow: least privilege, tamper resistance of the trusted computing base, complete mediation, secure information flow, and accounting for the human as a weak link. As evidence, the authors analyzed eleven real-world attacks on AI agents and mapped each to the principles it violated. The attacks included data exfiltration from the ChatGPT macOS app, a Claude Code exfiltration flaw, a Microsoft Copilot exfiltration vulnerability, and the AgentFlayer attack on Cursor through a malicious Jira ticket. Every one of the eleven violated the secure information flow principle, the paper said, while most violated the least privilege principle. The authors rejected the idea that stacking machine-learning guardrails amounts to a defense. “Merely stacking ML models does not constitute true defense-in-depth,” they wrote, because the guard models “often share the same statistical failure modes as the primary agents they monitor.” To put the principles into practice, the authors proposed three security mechanisms, each tied to an open research problem the community has yet to solve. The first is separating instructions from data, because language models mix the two in a single stream of tokens with no source-level distinction between them. The second is verifiable least-privilege policy generation, made difficult because security policies for agents are written in natural language and shift as a task evolves, which makes them hard to translate into rules a system can enforce. The third is information flow control, since tracking how sensitive data moves through a model remains unsolved. Beyond the model The paper challenges one of the dominant assumptions shaping enterprise AI-security efforts over the past two years, that increasingly capable models, alignment techniques, and prompt defenses would eventually make AI systems sufficiently secure for enterprise deployment. Instead, the researchers argue AI agents should increasingly be treated more like operating environments or distributed systems than conventional enterprise applications because they combine reasoning, autonomy, memory persistence, and external tool execution inside a single operational layer. “Security guarantees cannot emerge solely from better prompts, alignment tuning, or model-side mitigations,” the paper said, arguing enterprises instead need stronger runtime isolation, containment boundaries, least-privilege execution, and workflow observability controls around AI agents. That creates situations where prompt injection is no longer simply a content-manipulation issue but potentially a workflow-execution and systems-integrity problem capable of influencing downstream actions across interconnected enterprise environments. The visibility problem The researchers also argue that current enterprise security tooling lacks sufficient runtime visibility into how AI agents actually reason, invoke tools, retain memory, and execute actions across enterprise systems. Another paper published last week also points to a similar problem from a different angle, arguing that traditional endpoint detection and response platforms cannot adequately inspect AI-agent reasoning flows, prompt chains, memory interactions, or dynamic tool execution. The paper proposed what researchers described as an “agentic detection and response or ADR” framework designed specifically for AI-agent environments. “Current security tools are not designed to observe agent cognition or reasoning traces,” the researchers wrote, arguing that existing enterprise security stacks were built to monitor deterministic applications and endpoint activity, not systems capable of autonomous planning, probabilistic reasoning, and dynamic workflow orchestration. The paper described a production deployment monitoring more than 10,000 AI-agent sessions daily across roughly 7,200 hosts, where researchers said the framework identified hundreds of credential-exposure incidents and other agent-related risks spanning 26 attack categories. On a benchmark the team introduced, called ADR-Bench, the system detected 67% of attacks with zero false positives, outperforming three baselines, including Meta’s LlamaFirewall, by two to four times in F1-score, the paper said. On AgentDojo, a public prompt injection benchmark, it detected all attacks with three false alarms out of 93 tasks.
Oncology Institute Discloses Data Breach

www.securityweek.com · 2 months ago

The affected third-party vendor has not been named, but one possible candidate is TriZetto.
Ghost CMS CVE-2026-26980 Exploited to Hijack 700+ Sites for ClickFix Attacks

thehackernews.com · 2 months ago

Threat actors are exploiting a recently disclosed critical security flaw in Ghost CMS to inject malicious JavaScript code with an aim to fuel ClickFix attacks. According to QiAnXin XLab, the activity involves the exploitation of CVE-2026-26980 (CVSS score: 9.4), an SQL injection vulnerability in Ghost's Content API that could allow an unauthenticated attacker to read arbitrary data from the
Threat Intel: ShinyHunters Leaks 9.4GB Database of 7-Eleven Franchisee Systems Post-Extortion Refusal

www.technadu.com · 2 months ago

An extortion campaign by ShinyHunter exposed customer data belonging to over 185,000 accounts associated with 7-Eleven. The breach specifically compromised customer login credentials. Affected customers have been notified, and the company is currently investigating the full scope of the data leak.
APT Group Patches termsrv.dll to Enable Multiple RDP Sessions

gbhackers.com · 2 months ago

A sustained cyber espionage campaign attributed to the Cloud Atlas advanced persistent threat (APT) group has introduced a stealthy technique that modifies the Windows termsrv.dll library to enable multiple Remote Desktop Protocol (RDP) sessions on compromised systems. Observed throughout 2025 and continuing into 2026, the activity primarily targets government and commercial entities in Russia and
As AI speeds coding, CVE Lite CLI keeps security deliberately AI-free

www.csoonline.com · 2 months ago

As AI coding assistants accelerate software development, one OWASP-backed open-source project is arguing that dependency security tooling still arrives too late to be truly useful. CVE Lite CLI, a JavaScript and TypeScript dependency vulnerability scanner focused on local lockfile analysis, is positioning itself around a simple idea. Developers should see dependency risks while they are still writing code, not hours later inside a failing CI pipeline. “What developers are missing is early feedback at the point where the dependency decision is made,” Sonu Kapoor, creator and maintainer of the project, told CSO. According to Kapoor, traditional CI-centric workflows often disconnect developers from the dependency choices that introduced risk in the first place. CVE Lite CLI scans npm, pnpm, and Yarn lockfiles using OSV vulnerability data and claims to focus heavily on remediation guidance, including separating direct and transitive vulnerabilities, validating upgrade targets, and recommending actionable fix paths. The project is being pitched as a “local-first” developer tool, as opposed to a replacement for enterprise software composition analysis ( SCA ) platforms, much like how developers already use ESLint or unit tests locally before CI runs them again later. CVE Lite CLI targets an overlooked pain point CVE Lite CLI is essentially trying to solve a workflow problem, Kapoor says many developers quietly struggle with. Dependency security checks often arrive after the work is already done. The tool scans JavaScript and TypeScript lockfiles locally across npm, pnpm, and Yarn projects, so developers can understand dependency risk while they are still coding, not later in response to a failing CI pipeline. Instead of just focusing on detection, the tool claims to look into subsequent questions like whether the issue is direct or transitive, whether there is a clean upgrade path, or whether upgrading one package actually removes the vulnerable dependency. “In one real case, CVE Lite CLI skipped 27 package versions before finding a safer version to recommend,” Kapoor said, explaining the granularity of the tool. “That is the kind of work developers should not have to do manually by reading logs and retrying upgrades one by one.” Kapoor said the tool can be configured for JSON, SARIF, or HTML outputs and can also be integrated into CI workflows as a GitHub Action. AI could be making things worse The argument arrives as software supply chain security continues to collide with AI-assisted development practices that allow developers to generate code, integrate packages, and restructure projects much faster than before. Kapoor said this velocity changes the nature of dependency risk itself. “AI coding assistants have made this more important, not less,” he said. “That speed is useful, but it also means dependency decisions can happen quickly and sometimes without the same level of manual review. I do not think AI assistants remove the need for security checks.” On the contrary, they increase the need for fast, local, explainable checks that can be run while the work is happening, he added. One cited example involved scans against lint-staged, a widely used JavaScript tooling package. According to Kapoor, a standard “npm audit, omit=dev” workflow failed to surface a production dependency issue that CVE Lite CLI later identified through lockfile analysis. “Honestly, I don’t think most developers understand those blind spots in detail, and I do not mean that as criticism of developers,” he said. “The dependency graph in a modern JavaScript project is extremely noisy.” A developer meaning to install one direct dependency may end up with hundreds or thousands of transitive packages. CVE Lite CLI isn’t falling for AI The project also deliberately avoids turning itself into a broader AppSec platform, despite growing industry pressure to consolidate security tooling into an AI-enabled ecosystem. “I do think security tooling has become too heavy for the day-to-day developer workflow,” Kapoor said. “That does not mean those platforms are bad. It means they often serve security organizations better than they serve the individual developer trying to make a safe dependency decision during a normal coding session.” This philosophy also extends to the project’s approach toward AI itself. While CVE Lite CLI includes integrations that help AI coding assistants interpret scan results, Kapoor said the underlying vulnerability analysis intentionally remains deterministic. “I do not think AI should decide whether a CVE exists,” he said. “That part needs to be boring, repeatable, and auditable.” Instead, the project uses AI as what the founder described as an “explanation and workflow layer” around scan results rather than as the scanner itself. “CVE Lite CLI includes AI assistant skills that teach tools like Claude Code, Codex CLI, Gemini CLI, Cursor, and GitHub Copilot how to run CVE Lite CLI, read its structured output, and help the developer understand or prioritize the remediation plan,” Kapoor explained. Caution around expansion Kapoor said he has been receiving positive feedback from the companies and developers using CVE Lite CLI in real workflows, asking him whether the same approach could support.NET or Python ecosystems. “That interest is encouraging because it tells me the local-first, remediation-oriented model is resonating beyond the original JavaScript and TypeScript use case,” he said. “But I am cautious about expanding the current tool too broadly.” The explanation he gave was simple. Each ecosystem, he believes, has its own package manager behavior, lockfile format, dependency graph semantics, advisory sources, and remediation patterns. “Adding those directly into CVE Lite CLI could make the tool heavier and less clear for the JavaScript and TypeScript developers it was originally designed to help.” The project has now been adopted into the OWASP foundation ecosystem as an official OWASP project and is available for free to developers on GitHub.
The Alert Firehose Finally Meets Its Match

thehackernews.com · 2 months ago

Ask a cybersecurity pro about Network Detection and Response (NDR) and you might still hear "Noisy," "Too much data." But ask the teams running NDR that includes agentic AI capabilities and you'll hear they're actually using it to catch threats earlier, triage faster, and chase fewer false positives. The old complaint lingers in part because reputations are sticky, and because NDR has evolved
266,000 Affected by Data Breach at Radiology Associates of Richmond

www.securityweek.com · 2 months ago

Threat actors stole files containing names and protected health information from the healthcare organization’s systems.
Anthropic: Mythos Detected 23,000 Potential Vulnerabilities Across 1,000 OSS Projects

www.securityweek.com · 2 months ago

Many findings have been confirmed to be critical or high-severity vulnerabilities and the number will continue to increase.
Laravel-Lang Packages Poisoned for Malware Delivery

www.securityweek.com · 2 months ago

Published within a 15-minute window, the malicious tags introduced backdoors to exfiltrate CI secrets.