Live

Intelligence Feed

Latest threat intelligence articles from trusted security sources, auto-processed to extract entities, IoCs, and TTPs.

Filtered by source: hacker-news Clear filter
Cursor Says Stolen Fireworks AI Key Led to Trojanized Claude Code Extension Incident

2d ago · hacker-news

An attacker compromised Cursor's extension marketplace by exploiting a stolen access key originally issued to Fireworks AI, which had been exposed due to a prior breach. The attacker used the key to access a read-only storage bucket containing limited coding session traces, from which they extracted credentials to pivot into Cursor's internal systems. They then uploaded a trojanized version of Anthropic's official Claude Code extension (versions 2.1.312 and 2.1.313) to the marketplace, which was available for approximately 11 hours and potentially affected up to 5,000 users. The malicious extension attempted to install a persistent SSH backdoor, and the attacker also accessed archived source code repositories, customer support data, and user credentials. Cursor responded by deactivating the compromised key, rotating credentials, and taking the extension marketplace offline.

2 IoCs
Attackers Exploit UnrealIRCd Flaw That Enables Server-Level Command Injection

3d ago · hacker-news

A high-severity vulnerability, CVE-2026-114713, in UnrealIRCd versions prior to 6.2.7.1 is being actively exploited to enable server-level command injection. The flaw exists in the labeled-response module, which fails to properly escape IRCv3 'label' message tags, allowing untrusted users to inject and execute server-to-server commands with elevated privileges. Exploitation began shortly after the patch release on November 7, 2026, and CISA has added the vulnerability to its Known Exploited Vulnerabilities catalog, mandating federal agencies to remediate by November 16, 2026. The full scope and objectives of the ongoing attacks remain unclear, but the project maintainer has described it as the most critical issue in UnrealIRCd's history.

Hackers Exploit N-central Flaws to Infect 21 Servers, Including 19 Running Hotfix 2

4d ago · hacker-news

Threat actors are actively exploiting a vulnerability chain dubbed 4EverAble2Shell in N-able N-central, which combines a flaw in the Apache Axis component with a privilege escalation issue to achieve remote code execution as root. The exploit allows unauthenticated attackers to execute arbitrary OS commands via unsafe class instantiation during SOAP fault processing. Attackers have used the access to deploy a Monero cryptocurrency miner via a shell script downloaded from a malicious server. At least 21 servers have been infected, including 19 that had already applied Hotfix 2, indicating the hotfix alone is insufficient for full protection.

3 IoCs 1 Malware 1 CVEs
Hackers Exploit PHP2Shell Flaws in Check Point Gateways to Deploy TSGA Backdoor and Wipe ESXi Files

4d ago · hacker-news

An unknown threat actor exploited the PHP2Shell vulnerabilities (CVE-2026-109618, CVE-2026-109619) in unpatched Check Point Security Gateways to gain initial access, then used a local privilege escalation flaw (CVE-2026-123039) in Gaia OS to deploy the TSGA backdoor and establish persistent access. The attackers deployed PHP web shells, moved laterally to VMware ESXi hosts via SSH, and executed the destroy_v6 wiper malware to delete files and leave politically themed notes. The campaign targeted Israeli organizations initially, then expanded globally, with some evidence suggesting possible ties to the China-nexus group Armory Castle, though attribution remains uncertain due to the use of wiper malware and political messaging.

4 IoCs 2 CVEs
Operation Endgame Takes Down Gootloader, CastleRAT and StealC Servers, Seizes 6.4M Stolen Logins

4d ago · hacker-news

Operation Endgame, an international law enforcement effort, dismantled infrastructure associated with Gootloader, CastleRAT, and StealC—three malware tools used in cybercrime operations. Gootloader, a loader malware, spread via SEO-poisoned fake legal documents on compromised WordPress sites, enabling ransomware attacks and credential theft. CastleRAT is a remote access trojan allowing full control over infected systems, while StealC is an infostealer that exfiltrates login credentials. Authorities seized 235 servers and 257 domains across the three campaigns, along with 6.4 million stolen credentials, though no arrests were announced. The takedown included disruption of bulletproof hosting provider Virtualine/Omegatech, which supported StealC's infrastructure.

1 IoCs 3 Malware
SilentAir Hijacks Asian Telecom Networks to Intercept SMS 2FA Codes and Spy on Taiwanese Users

4d ago · hacker-news

SilentAir is a newly identified cyber espionage group assessed by Enea to be a state-level actor, likely operating with ties to China, that has hijacked telecom infrastructure in at least six Asian countries to intercept SMS two-factor authentication codes and conduct surveillance. The group exploited SS7 and Diameter signaling protocols via compromised equipment within mobile operators' networks, enabling them to redirect calls and texts by spoofing location update messages. SilentAir conducted large-scale silent SMS operations, primarily targeting users in Taiwan, using disguised messages to verify interception success without alerting victims. The group's activity aligns with China Standard Time and overlaps temporally with known Liminal Panda intrusions, though definitive attribution remains unconfirmed.

1 Actors 1 Malware
Trezor Confirms Hidden Implants in CryptoBilis-Sold Wallets That Can Steal Recovery Phrases

4d ago · hacker-news

Trezor confirmed that three of its Safe 5 hardware wallets purchased from the reseller CryptoBilis contain hidden malicious implants designed to steal recovery phrases. The implant, positioned between the main board and screen, captures the recovery phrase during setup or restoration and transmits it over mobile networks to attacker-controlled servers. Although no fund losses have been reported, Trezor advises users who bought from CryptoBilis since March to move funds to a new wallet with a new recovery phrase and warns against using or restoring backups on potentially compromised devices.

FBI Says North Korean Hackers Stole Over $400 Million in Crypto Using Deepfake Video Calls

4d ago · hacker-news

The FBI has attributed a social engineering campaign named FaceStealer to North Korea-linked threat actors, specifically the BlueNoroff group, which has stolen over $400 million in cryptocurrency over the past 18 months. The campaign uses deepfake video calls and surreptitiously recorded videos to impersonate trusted contacts during fake Zoom or Teams meetings, tricking high-value targets—primarily cryptocurrency and Web3 industry professionals—into clicking malicious links or running scripts that install malware. The attack chain leverages compromised Telegram accounts to propagate, using typosquatted domains and phishing kits to deliver payloads, ultimately enabling credential theft, supply chain compromises, and lateral movement. A related but distinct campaign, Contagious Interview, uses fake job interviews and weaponized code repositories to deploy malware such as BeaverTail and InvisibleFerret.

2 IoCs 1 Actors 2 Malware
New Mior and Skyfall Attacks Let Rogue Cloud Hosts Steal Secrets From AMD SEV-SNP VMs

4d ago · hacker-news

Two new attacks, Mior and Skyfall, have been disclosed that compromise AMD SEV-SNP-protected virtual machines by exploiting flaws in the processor's startup and transient execution behaviors. Mior allows a malicious host with control of the hypervisor and BIOS to manipulate the Reverse Map Table (RMP) during VM boot, enabling unauthorized memory access and falsified attestation reports. Skyfall leverages transient scheduler attacks (TSA) to leak VM register data and disk-encryption keys from failed VM startups, even without hardware tampering, by reading leftover cached state through timing side channels.

PostgreSQL Patches 52 Flaws, Including 22-Year-Old Bug Enabling Authenticated RCE

4d ago · hacker-news

PostgreSQL has patched 52 security flaws, including a critical type confusion vulnerability (CVE-2026-6460) that allows authenticated remote code execution with a CVSS score of 8.8. The flaw, present for 22 years, affects all versions prior to 18.7, 17.12, 16.16, 15.20, and 14.25, and can be exploited by a database user with object creation privileges to execute code with PostgreSQL's operating system user rights. Security researcher Mehmet Ince demonstrated a working exploit that crashes the server and leaks memory addresses, though the final command execution payload was not disclosed. Administrators are advised to update immediately and rebuild C-language extensions, as no workaround exists for unpatched systems.

Calendar Invites and Live Proxies: Inside a Google Session Hijacking Campaign

5d ago · hacker-news

Since mid-August 2026, a phishing campaign has been hijacking Google sessions by leveraging malicious Google Calendar invites sent from compromised accounts. The lures appear legitimate, often using BCC and familiar sender domains, and redirect through Google's own URL shortener or trusted services like share.google.com. Upon clicking, victims are proxied through a live reverse proxy framework called Nyasher, which captures credentials and multi-factor authentication tokens in real time by relaying them to the legitimate Google sign-in page. The infrastructure uses ephemeral domains, wildcard certificates, and client-side JavaScript rendering to evade detection, with operators rotating tactics to bypass static and machine learning-based defenses. The campaign, dubbed Operation CloudFire by Material Security, was mitigated through infrastructure takedowns via Cloudflare abuse reporting, protecting broader user bases beyond individual organizations.

2 IoCs
New N-central Exploits Work Against Patched Servers, Enable Root RCE and Windows SYSTEM Access

5d ago · hacker-news

New exploits named 4EverAble2Shell and Agent2Shell have been disclosed against N-central on-premise servers, enabling unauthenticated remote code execution with root privileges and command execution as SYSTEM on Windows agents, respectively. These vulnerabilities affect patched versions, including build 2026.4.1.26, and rely on the non-default Legacy API being enabled. Probing activity has been observed from IP 157.20.241[.]16, leveraging the Apache Axis flaw CVE-2026-90436 for initial access, followed by attempts to exploit 4EverAble2Shell and download a malicious shell script.

1 IoCs 1 CVEs
Google Cloud Warns Three AMD Turin CPU Flaws Could Leak Data Across Processes in VMs

5d ago · hacker-news

Google Cloud has issued a security advisory regarding three CPU side-channel vulnerabilities in AMD's Turin (fifth-gen EPYC) processors that could allow data leakage across processes within virtual machines. The flaws are classified as Transient Scheduler Attacks (TSA): TSA-SQ (CVE-2024-36350), TSA-L1 (CVE-2024-36357), and a newly disclosed variant TSA-Fill (CVE-2026-0004). These vulnerabilities may enable an attacker with code execution on the same core to infer sensitive data from other processes or the kernel via the L1 cache. Google has applied host-level mitigations but requires customers to patch guest operating systems and perform a full stop-and-start of VMs to fully mitigate the risk.

2 CVEs
SCKit Malware Turns ScreenConnect Support Sessions Into Potential MSP Attack Paths

5d ago · hacker-news

The SCKit malware campaign exploits CVE-2026-84869, a critical improper privilege management flaw in ConnectWise ScreenConnect, to execute unauthorized file transfers and code execution during active remote sessions. The malware uses legitimate ScreenConnect features like Run Command to deploy a malicious MSI installer and PowerShell scripts, enabling persistence and lateral movement by hijacking authenticated technician sessions. A key component, RelayTrap, enables reverse-pivoting to infect technician workstations, leveraging trusted remote support functionality to propagate across managed service provider (MSP) environments without requiring new vulnerabilities once initial access is achieved.

5 IoCs
Citrix Patches 12 NetScaler and Secure Access Client Flaws, Including Authentication Bypass

5d ago · hacker-news

Citrix has patched 12 critical vulnerabilities across its NetScaler ADC, NetScaler Gateway, NetScaler Console, and Secure Access Client products. The flaws include multiple memory overflow and memory corruption vulnerabilities that could lead to denial-of-service conditions, as well as improper authentication and privilege management issues that could allow unauthenticated access or privilege escalation. The most severe vulnerability, CVE-2026-107412, has a CVSS score of 9.5. Citrix urges customers to apply updates despite no known active exploitation at the time of disclosure.

1 CVEs
Critical Red Hat Quay Flaw Lets Attackers Read and Overwrite All Private Container Images in a Registry

5d ago · hacker-news

Two critical vulnerabilities in Red Hat Quay have been disclosed, allowing attackers to read and overwrite private container images in any repository. CVE-2026-123710 enables an unauthenticated attacker to add a trusted signing key with no expiry, allowing them to generate valid access tokens for any repository, even without credentials. CVE-2026-110697 allows a user who can create a namespace with an underscore to exploit an SQL injection in the robot account listing logic, stealing other tenants' robot account tokens and gaining unauthorized access to their images.

New SignDrop Zero-Day Exploit Turns Windows CLFS Security Feature Into SYSTEM Access

5d ago · hacker-news

A zero-day privilege escalation exploit named SignDrop targets the Windows Common Log File System (CLFS) driver's log-authentication feature, allowing a local user to gain SYSTEM privileges. The exploit abuses the creation of companion .cnpf files by redirecting their path via symbolic links when a privileged process, such as Windows Defender, opens a prepared log. This results in a SYSTEM-owned file that the attacker can write to, enabling escalation to a full SYSTEM shell. The vulnerability affects systems with CLFS authentication mode set to 0 or 1, and while no official patch or CVE has been issued, the exploit has been publicly demonstrated and independently verified to work, though with reliability issues.

3 IoCs
ThreatsDay: 59 Linux Kernel Bugs, Falcon Spyware, GSocket Backdoors + 17 More Stories

5d ago · hacker-news

Multiple active threat campaigns were observed in November 2026, including North Korean actors laundering $640 million stolen from Binance through complex blockchain transactions and meme coins. Microsoft reported a surge in GSocket backdoor usage on Linux servers, enabling persistent access via public relay networks. Falcon browser-based spyware, linked to PeckBirdy, was used to steal credentials and session data from government webmail portals in the Middle East and Southeast Asia. A new ransomware group, Citrus, leveraged AI-generated scripts and contractor access to deploy Sn00py-family ransomware. Additionally, Proofpoint identified BluePhishing, a new OAuth abuse technique targeting Microsoft Entra AI agent consent. RevStealer malware expanded its distribution through over 100 domains and trojanized AI tools, stealing credentials and cryptocurrency data.

6 IoCs 1 Actors 1 Malware 3 CVEs
Railway Says Stolen Customer AI Keys Led to $740,000 in Estimated Unauthorized Usage Charges

6d ago · hacker-news

In October 2026, Railway suffered a data breach in which an attacker exploited a previously unknown flaw in its build system to escape filesystem isolation and obtain internal credentials. Using these, the attacker accessed a production database and decrypted environment variables from approximately 100,000 workspaces, specifically targeting AI provider API keys. The stolen keys were used to generate an estimated $740,000 in unauthorized AI service usage. The attacker also obtained GitHub OAuth tokens and cloud credentials, though no customer database access or code repository exfiltration was confirmed. Railway has taken responsibility for covering costs where providers do not issue refunds and has since hardened its systems, including isolating builds in short-lived VMs and rotating all potentially exposed credentials.

Microsoft Patches 1,259 Flaws, Including Exploited Outlook Bug That May Expose NTLMv2 Hashes

6d ago · hacker-news

Microsoft released patches for 1,259 vulnerabilities, including an actively exploited Moderate-severity flaw in Microsoft Outlook (CVE-2026-124194) that allows spoofing and potential exposure of NTLMv2 hashes. An unauthenticated attacker can send a specially crafted message file that, when opened, causes Outlook to connect to an attacker-controlled network location without warning, potentially disclosing authentication credentials. The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added the vulnerability to its Known Exploited Vulnerabilities catalog, requiring federal agencies to patch by November 24, 2026. Other notable vulnerabilities include remote code execution flaws in Windows Search, LDAP, Schannel, and Windows Xbox Live Networking Service, as well as a privilege escalation flaw in Azure OpenAI.

eMudhra Revokes 8 More CAs It Previously Declared Unaffected by Key Compromise

6d ago · hacker-news

eMudhra revoked 8 additional intermediate certificate authorities (CAs) due to a flawed internal Python script used during a key ceremony on April 30, 2025. The script improperly handled private keys by extracting them from hardware security modules (HSMs) into memory and, in the case of ECDSA keys, truncating them to a single byte, resulting in weakened cryptographic strength. Although the RSA keys were not truncated, their private components were exposed in memory, violating internal control policies and preventing eMudhra from proving compliance with its own certificate practice statement. The company had previously declared these CAs unaffected but reversed course after further review, ultimately revoking them due to inability to demonstrate proper key handling, with no known exploitation reported.

AntiFox AI Pentesting Platform Linked to Attacks That Accessed AWS Production Secrets

6d ago · hacker-news

Cybersecurity researchers have identified malicious use of the AI-driven penetration testing platform AntiFox, which has been leveraged in attacks to steal AWS credentials and access production secrets from compromised environments. The attacks begin with exploitation of internet-facing IIS applications, leading to the deployment of ASPX web shells via the w3wp.exe process, enabling command execution and credential harvesting. The threat actors then abuse valid AWS IAM identities to enumerate and access cloud services including S3, Secrets Manager, and EC2, with confirmed successful GetSecretValue calls retrieving production secrets. The campaign highlights the growing abuse of autonomous AI-powered pentesting tools like AntiFox and ARTEX by threat actors to conduct rapid, scalable attacks with minimal human intervention.

1 IoCs
Critical Zoom Workplace Flaw Could Allow Unauthenticated Remote Privilege Escalation

6d ago · hacker-news

Zoom has patched two security vulnerabilities in Zoom Workplace for Windows that could allow unauthenticated remote privilege escalation. The critical flaw, CVE-2026-108186, stems from improper input validation and affects versions from 7.1.0 to before 7.1.8. The second vulnerability, CVE-2026-108187, involves origin validation and impacts versions prior to 7.1.5. Exploitation could enable attackers to escalate privileges without authentication, though there is no evidence of in-the-wild exploitation to date.

Apple Patches iMessage Group Chat Privacy Flaws and Bluetooth Pairing Vulnerabilities

1w ago · hacker-news

Apple has released security updates to address multiple vulnerabilities, including two in iMessage that could allow a remote attacker to join a group chat and access messages sent prior to their addition. The vulnerabilities, tracked as CVE-2026-113961 and CVE-2026-113960, stem from authorization issues and were fixed through improved validation. Additionally, a Brotli compression library out-of-bounds write vulnerability (CVE-2026-111497) was patched, which could allow memory corruption from malicious input, and several Bluetooth and AirPlay flaws were resolved that could enable unauthorized pairing or memory corruption.

1 CVEs
Binance Warns of iOS Sandbox Escape That Could Expose Crypto Wallet Keys, No Fix Confirmed

1w ago · hacker-news

A sandbox escape vulnerability in iOS 26 and 27, dubbed 'worse_query', has been publicly released on GitHub by researcher forcequit in collaboration with 0xjohnny. This exploit allows malicious apps to read and modify data from other apps' containers without requiring jailbreak or kernel exploits, potentially exposing cryptocurrency wallet seed phrases and private keys. A file manager app called WQ Files, developed by Christoph Kolbicz and built on this exploit, demonstrates the risk through sideloaded installation via tools like AltStore and SideStore. Apple's recent iOS updates (26.7.3 and 27.2.1) do not appear to address the issue, and no official fix has been confirmed.

5 IoCs
Fireworks AI Confirms Wider Breach Affecting Customer Datasets, Model Weights and Notion AI Chats

1w ago · hacker-news

Fireworks AI confirmed a data breach that began with unauthorized access to internal credentials, leading to broader compromise of customer datasets, model weights, and billing information. Attackers accessed the company's development environment via a third-party cloud provider and ran unauthorized workloads between October 23 and October 25, 2026. The breach also impacted Notion AI, exposing chat transcripts from the Insights feature for fewer than 1% of workspaces. Fireworks AI attributed the activity to a financially motivated threat actor group conducting 'LLMjacking'—stealing access to large language models to resell illicitly.

Blind Eagle Uses Fake Court Notices to Deliver Remote Access Trojan Across Latin America

1w ago · hacker-news

The threat actor Blind Eagle, also known as APT-C-36, has launched a new campaign targeting organizations in Latin America using phishing emails with judicial and administrative-themed lures in Portuguese and Spanish. These emails deliver a malicious attachment containing a weaponized SVG file that performs HTML smuggling to deploy a JavaScript dropper, which then executes a Visual Basic Script to bypass AMSI and ETW while launching an AsyncRAT-based remote access trojan. The malware establishes persistence via the Windows Startup folder, conducts environment reconnaissance, and targets sensitive data in judicial, fiscal, and educational sectors, with a focus on Colombia.

1 Actors 2 Malware
Attackers Exploit AhsayCBS Flaws to Deploy Sandbox-Evading Python Backdoor

1w ago · hacker-news

Threat actors exploited two zero-day vulnerabilities, CVE-2026-105133 and CVE-2026-105134, in AhsayCBS to gain unauthorized access and execute arbitrary commands on backup servers. The attackers deployed multiple JSP web shells, cryptocurrency miners, and a custom Python-based backdoor designed to evade sandbox analysis by checking the host's MachineGuid and presence of a marker file. The backdoor loads an in-memory x64 DLL that communicates with a C2 server, while the attackers also deployed tunneling tools and remote access software to maintain persistence and potentially harvest credentials from backup systems, including Veeam servers.

6 IoCs
Hackers Exploit Critical Xen Orchestra Auth Bypass Flaw to Gain Admin Access

1w ago · hacker-news

A critical authentication bypass vulnerability in Xen Orchestra, tracked as VSA-2026-062 and not yet assigned a CVE, is being actively exploited in the wild. The flaw allows attackers to bypass authentication by sending a specially crafted cookie (authenticationToken=j:%7B%7D), leading to admin-level access and arbitrary code execution on managed XCP-ng nodes. Malicious activity has been linked to an IP address and a TOR domain used as a C2 server for RedLine Stealer. Indicators include specific virtual machine names, cron files, and configuration parameters. Patches are available in Xen Orchestra 6.9.1 and 6.10.0, and xo-server versions 5.211.2 and 5.213.1.

6 IoCs 1 Malware
SGLang RCE Exploit Breaches NEAR AI, Steals Data From 18 Agents and About 100 Users

6d ago · hacker-news

An attacker exploited a remote code execution vulnerability in the open-source SGLang inference engine to breach NEAR AI's infrastructure, gaining access to an inference server running within a trusted execution environment (TEE). Despite the TEE's design to protect confidential data, the flaw in SGLang allowed the attacker to access processed requests, including those in confidential mode. The attacker exfiltrated account details of about 100 users from a deprecated internal system called Private Chat and accessed data from 18 AI agents on an older non-TEE deployment, including chat history, settings, logs, and two exposed NEAR AI Cloud API keys, which have since been revoked.

Next →