Skip to main content
Mallory
LiveUpdated continuously · 8 new this hour

Today in threat intelligence.

Mallory correlates nearly 9,000 sources and methods: vendor advisories, government alerts, researcher write-ups, underground collection, scanning and honeypots. We extract the facts that matter, so you can act on them right now.

nearly 9,000
sources monitored
continuously
20,006
reports analyzed
last 24 hours
612
stories surfaced
last 24 hours
Free daily digest

Your morning threat briefing.

The day's top threat intelligence stories, plus everything else filed in the last 24 hours. One email each morning.

Get the daily digestFree · Unsubscribe anytime
Top storyInternet Facing Service VulnerabilityPrivilege Escalation TechniqueOpen Source Dependency Vulnerability

Critical NGINX Flaws Enable Unauthenticated RCE; Researchers Demonstrate Root Escalation Chain

Patch & Detection Intelligence · 114 sources · Updated 40min ago

Researchers disclosed critical vulnerabilities in NGINX Open Source and NGINX Plus that could allow unauthenticated attackers to execute code on exposed servers. Rapid7 and Wiz describe CVE-2026-123007, a CVSS v4.0 9.2 HTTP/2 flow-control integer underflow that can cause information disclosure, memory corruption and code execution with worker-process privileges, including on static-file servers. Wiz also reports CVE-2026-123633, a similarly rated HTTP/2 connection use-after-free affecting reverse-proxy processing. Xint describes five independent NGINX flaws across HTTP/2, resolver, upstream-proxying, HTTP/3 and mail-proxy configurations, collectively presented as nginx2root alongside a separate Linux SCTP privilege-escalation flaw. Xint demonstrated chaining worker-level execution with the SCTP vulnerability to obtain host root privileges, including from a Docker container. Croatia’s CERT separately warns of four critical flaws, also naming CVE-2026-121810 and CVE-2026-123634. The reports describe laboratory exploitation, not confirmed attacks in the wild; Rapid7’s demonstrated HTTP/2 exploit required approximately 128 GB of traffic and a specific glibc memory-copy implementation.

Latest developments
  1. ~15h agoSysdig validates SCTP mitigation and tests NGINX post-exploitation detections
  2. ~203d agoCSIRT Panamá issues NGINX2Root advisory and remediation guidance

Latest

State Sponsored EspionageActively Exploited Vulnerability

ARMORY CASTLE Exploits VMware vCenter Flaw in Espionage Intrusion

Active Exploitation · 4 sources · Updated 1h ago

Google Threat Intelligence Group attributed an intrusion involving exploitation of VMware vCenter vulnerability CVE-2026-123890 to ARMORY CASTLE, formerly UNC5221, a suspected PRC-nexus espionage actor targeting virtualization and security infrastructure. The report characterized BONESAW as an expansion into post-compromise zero-day exploitation and described specialized tooling for vCenter identity compromise, virtual disk access, and persistence, including SAWSNAKE and vmfs_browser.py.

Botnet InfrastructureCybercrime Service Ecosystem

BADBOX 3.0 Disrupted After Generating 40 Billion Fraudulent Ad Bid Requests Daily

Malware Intelligence · 3 sources · Updated 1h ago

HUMAN Security’s Satori team identified and disrupted BADBOX 3.0, an operation that used compromised low-cost Android streaming boxes and fabricated traffic to generate roughly 40 billion advertising bid requests daily at its peak. The campaign ran hidden video ads, impersonated legitimate connected-TV apps, and enrolled devices in residential proxy networks without owners’ consent. Its traffic exceeded 4% of connected-TV traffic observed by the Human Defense Platform. Findings published November 18 linked the operation’s software catalog to Chinese electronics manufacturer TXCZ through infrastructure and device-specific evidence. Researchers also found Vo1d botnet code and assessed that the PXYEDGE proxy network succeeded IpMoYu. The findings did not establish a unique infected-device count; bid requests do not equate to ads watched or payments made.

Internet Facing Service VulnerabilityWidely Deployed Product Advisory

Drupal Patches Critical Installer Flaw That Could Allow Site Takeover

Attack Surface Exposure · 5 sources · Updated 42min ago

Drupal has patched two core vulnerabilities, including CVE-2026-132165, a critical installer flaw that could allow unauthenticated attackers to reinstall the CMS, potentially leading to website takeover or remote code execution. The second flaw, CVE-2026-132166, is a moderately critical gadget-chain vulnerability that could enable remote code execution by attackers with administrative privileges when combined with another flaw permitting deserialization of untrusted data. Successful attacks could expose, alter, or delete data and allow malicious content to be inserted into affected websites.

Latest by section

Actively Exploited VulnerabilityOpen Source Dependency Vulnerability

PHP Plans Critical Security Fixes as PHP2Shell Campaign Compromises Hundreds of Hosts

Active Exploitation · 30 sources · Updated 34min ago

PHP has scheduled security updates for all supported versions on 19 November 2026, between 14:00 and 14:30 UTC, warning that some vulnerabilities are critical in nondefault configurations or uncommon setups. Full technical details, the complete CVE list and fixed version numbers remain unpublished ahead of release to limit attackers’ opportunities. Aikido reports that the updates are expected to address CVE-2026-115560, a remote code execution vector involving session upload progress and runtime changes to session serialization; its affected versions and severity score remain undisclosed. Administrators should prepare to deploy the updates immediately. For this specific issue, Aikido recommends disabling session upload progress before request parsing, restarting every PHP worker and retaining equivalent mitigations previously applied for PHP2Shell.

Actively Exploited VulnerabilityGovernment Vulnerability Catalog

CISA Flags Exploited Code-Execution Flaws in Apache Axis, Synology, LiteLLM and REDCap

Active Exploitation · 10 sources · Updated 2h ago

CISA added four actively exploited vulnerabilities affecting Apache Axis, Synology DSM/BeeStation OS, BerriAI LiteLLM and Vanderbilt University REDCap to its Known Exploited Vulnerabilities catalog. All four can enable arbitrary code execution, although exploitation prerequisites differ. Most are rated critical at CVSS 9.8; the LiteLLM flaw is rated high at 8.8 and requires proxy administrator credentials under the default configuration. Separately, Apache Roller developers reported active exploitation of a vulnerability in their software. CISA has not disclosed the nature or scope of the confirmed attacks, and ransomware involvement is listed as unknown for all four catalog additions.

Rapid WeaponizationBotnet Infrastructure

Germany’s BSI Rates Cyber Threats “Tense to Critical” as Exploitation Times Shrink

Active Exploitation · 3 sources · Updated 3h ago

Germany’s Federal Office for Information Security (BSI) has assessed the cybersecurity situation as “tense to critical,” warning that expanding attack surfaces and faster vulnerability exploitation are outpacing organizational defenses. The average interval between disclosure and first known exploitation fell from 53 days in July 2024–June 2025 to 25 days in July 2025–June 2026, reaching just eight days in the second quarter of 2026; some flaws were exploited within hours. BSI cited increasingly capable AI tools and geopolitical pressures as contributing factors, but did not attribute the deterioration to widespread, fully automated AI attacks. Small and medium-sized businesses and municipalities remain particularly exposed.

Actively Exploited VulnerabilityOpen Source Dependency Vulnerability

Apache Axis RCE Exploited in Confluence; N-central Attacks Linked to Same Flaw

Active Exploitation · 17 sources · Updated 3h ago

Rapid7 detected exploitation of CVE-2026-90436 against Atlassian Confluence Data Center on November 13, following Apache’s November 6 disclosure of the critical Apache Axis vulnerability. N-able separately reported exploitation of N-central strongly correlated with the same flaw. Exposure depends on non-default features: Confluence’s legacy XML-RPC/SOAP remote API or N-central’s Legacy API must be enabled. Confluence versions 1.x through 9.x are potentially affected; versions 10.x and later are unaffected because Atlassian removed the feature. CISA assigns the vulnerability a CVSS score of 9.8, while Atlassian rates its Confluence impact at 9.3 under CVSS v4.0.

Internet Facing Service VulnerabilityOpen Source Dependency Vulnerability

Critical PHP RCE Flaws Prompt Patching as Scanner Fixes Expose False-Safe Results

Attack Surface Exposure · 13 sources · Updated 3h ago

Two critical PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619, can enable unauthenticated remote code execution in web-serving installations, according to Red Sift. Both carry reported CVSS v4.0 scores of 9.2 and involve out-of-memory corruption and stale per-request state. Apache mod_php is identified as the highest-risk deployment because a demonstrated exploit runs during request parsing, before application code executes. The PHP security team reportedly released fixes in PHP 8.5.12, 8.4.27, 8.3.36, and 8.2.35. Organizations should inventory standalone and embedded PHP installations, apply fixed releases or vendor backports, restart all workers, and investigate potentially compromised systems; configuration mitigations do not replace patching.

Open Source Dependency VulnerabilityInternet Facing Service Vulnerability

Ruby2Shell Heap Overflow Exposes Rails and Sinatra Apps to Unauthenticated RCE

Attack Surface Exposure · 9 sources · Updated 6h ago

Ruby has patched Ruby2Shell (GHSA-34m4-m9fh-gqp5), a critical one-byte heap buffer overflow in string case conversion that can enable unauthenticated remote code execution. The flaw affects processing of certain legacy encodings; UTF-8 strings follow an unaffected code path. Veria Labs reports that default Ruby on Rails and classic-style Sinatra applications can expose the vulnerable functionality without authentication. Ruby maintainers are aware of proof-of-concept exploits achieving code execution in some scenarios, but sources report no known exploitation in the wild as of November 19, 2026. Researchers plan to publish their exploit in approximately one week.

Open Source Dependency VulnerabilityProof Of Concept Release

Ruby Fixes Seven Memory-Safety Flaws, Five With Proof-of-Concept RCE

Attack Surface Exposure · 6 sources · Updated 7h ago

Ruby maintainers released 4.0.10, 3.4.13, and 3.3.14 to fix seven memory-safety vulnerabilities in regular expression processing and string case conversion: three critical, three high, and one moderate. Five have proof-of-concept exploits demonstrating remote code execution, one has demonstrated attacker-controlled memory corruption, and another can disclose memory or crash a process. The advisories report no attacks exploiting these flaws. Exposure depends on how applications process attacker-controlled patterns or strings; one scenario involves a fixed regular expression with a case-insensitive backreference. CVE identifiers had been requested but were not yet assigned at publication.

Persistence MethodBotnet Infrastructure

Evooo1Bot Uses Five-Minute Cron Jobs to Restore Malware on Network Appliances

Attack Surface Exposure · 3 sources · Updated 21h ago

Evooo1Bot uses malicious cron jobs to restore malware on compromised Linux-based firewalls, routers, and other network appliances, undermining cleanup that only kills processes or deletes binaries. The botnet targets internet-facing devices through known vulnerabilities and SSH password guessing. Its persistence command installs a scheduled task that downloads and executes a loader every five minutes, alongside systemd, init, and login startup mechanisms. Evooo1Bot combines Mirai-derived attack code with encrypted command-and-control traffic and SOCKS proxy functionality that can conceal attacker activity and support movement deeper into networks.

Vendor Distribution CompromiseCommunication Services Sector Threat

Poisoned Third-Party JavaScript Exposes Major Websites to Skimmers and iPhone Exploit Loaders

Attack Tradecraft · 124 sources · Updated 3h ago

Researchers identified malicious third-party JavaScript reaching major websites through Optimizely, Kissmetrics, StorifyMe and minute-ly endpoints. Hudson Rock and researcher Kirk from Ohey linked 21 named web properties to affected resources, including BBC and Zoom. Payloads supported cryptocurrency-address substitution, payment-card skimming and conditional browser-data collection, alongside delivery infrastructure associated with the Coruna and DarkSword iPhone exploit kits. BBC and HSBC UAE scripts contained complete embedded malicious payloads, meaning blocking external attacker domains alone would not eliminate the threat. Archived scans showed mobile-routing and exploit-entry responses, but did not establish successful exploitation of an actual iPhone. The operators, initial access methods, full exposure window and confirmed theft or device-compromise totals remain unknown. [Hudson Rock](https://www.hudsonrock.com/blog/poisoned-scripts-on-bbc-zoom-and-other-major-websites-linked-to-iphone-exploit-loaders)

Search Ad ManipulationIdentity Impersonation Fraud

Malicious Google Ads Push Fake Developer Tools Through Suspected Advertiser Account Abuse

Attack Tradecraft · 2 sources · Updated 5h ago

Malicious Google search advertisements impersonated Codex, Claude Code, and Homebrew to lure developers into running unsafe installation commands, Cato Networks reported. Shared downstream domains and encrypted loaders linked parts of the Claude and Codex activity to an Atomic macOS Stealer (AMOS)-associated delivery system, without establishing a single operator or confirming every final payload. Other investigated delivery chains used Claude-themed JavaScript for Automation and a malicious macOS application named DayNote distributed through Homebrew lookalikes. Advertiser identity mismatches suggested account or identity abuse, while selective redirects were consistent with traffic cloaking; neither finding established compromise of the legitimate tool providers.

Enforcement ActionPhishing Campaign Intelligence

Scattered Spider Leader Sentenced to Six Years for Hacking and $11.6M Crypto Theft

Attack Tradecraft · 13 sources · Updated 5h ago

Tyler Robert Buchanan, a British leader of the Scattered Spider cybercrime collective, was sentenced to six years in a U.S. prison for cryptocurrency theft and conspiring to hack a dozen companies. Between September 2021 and April 2023, Buchanan and his accomplices used SMS phishing to steal employee credentials and personal information, then leveraged stolen information in SIM-swapping attacks to compromise cryptocurrency accounts and wallets. The scheme stole more than $11.6 million in virtual currency; four other conspirators also received prison sentences.

Phishing Campaign IntelligenceCredential Access Method

Microsoft 365 Device-Code Phishing Uses Fake SharePoint Lures to Gain Access Tokens

Attack Tradecraft · 2 sources · Updated 7h ago

Phishing campaigns analyzed by Hornetsecurity’s Threat Intelligence Lab used payment-themed emails, malicious HTML/SVG attachments and counterfeit SharePoint workspaces to target Microsoft 365 accounts. Obfuscated attachments redirected recipients to hosted lures that supplied verification codes and document-access instructions, steering them into Microsoft’s legitimate device-authorization flow. When victims signed in and completed any required MFA, they authorized an attacker-controlled client to receive access tokens rather than simply opening a document. The technique did not require stealing passwords or technically defeating MFA; potential access to corporate email and files depended on the permissions granted.

Government Diplomatic ThreatOperational Disruption

European Commission Cyberattack Disrupts HR, Reimbursement and Translation Services

Breach & Incident Intelligence · 11 sources · Updated 34min ago

A cyberattack disrupted European Commission internal services for more than 10 days, affecting personnel platform Sysper, reimbursement system MyPMO and translation applications, according to reporting attributed to POLITICO. Some employees were unable to complete routine personnel and reimbursement tasks. Commission spokesperson Thomas Regnier confirmed a campaign targeting multiple systems. Most affected applications had returned online, and an EU official said the Commission believed it had contained most of the attack, but the attackers, initial access method and full impact remained unknown. Investigators had not established whether personal or medical information was accessed or stolen.

Operational DisruptionInformation Technology Sector Threat

N-able Server Breach Disrupts Mail Assure and SpamExperts Management Across Three Regions

Breach & Incident Intelligence · 7 sources · Updated 4h ago

N-able confirmed that unidentified attackers gained unauthorized access to one server supporting its Mail Assure and SpamExperts Hosted Cloud services, disrupting email-security management across the Americas, Europe, and Asia-Pacific. The disruption began on November 16, 2026, and was initially attributed to platform configuration and service dependencies before being identified as a security incident. Management dashboards, APIs, quarantine access, and logging became unavailable, while core mail flow and inbound and outbound filtering remained operational.

Breach Disclosure NotificationIndustrials Sector Threat

Employer Solutions Staffing Group Breach Reportedly Affects Over 622,000 People

Breach & Incident Intelligence · 2 sources · Updated 9h ago

Employer Solutions Staffing Group (ESSG), a workforce management company based in Edina, Minnesota, disclosed a data breach estimated by BreachTally to affect 622,126 people nationwide. ESSG detected suspicious network activity on January 13, 2026, and determined that files were accessed or acquired without authorization that day. A review completed on October 13 identified personal information in the affected files. Reporting by teiss lists names, Social Security numbers, driver’s license numbers, financial account and payment card numbers, medical information, and health insurance data as exposed. However, the November 12 notification letter filed with South Carolina identifies names but leaves the additional data categories as an unfilled template field, limiting confirmation from that notice.

Breach Disclosure NotificationMass Credential Exposure

TfL Oyster Photocard Breach Exposes Personal Data of About 265,000 Customers

Breach & Incident Intelligence · 8 sources · Updated 12h ago

Transport for London (TfL) reported a cyber breach affecting approximately 265,000 Oyster photocard customers, exposing names, email and home addresses, phone numbers, dates of birth, and account security questions and answers. TfL detected suspicious activity on November 16 after customers received unexpected account-creation emails generated through its systems, and took the photocard website offline. TfL said the incident was contained and its current assessment found no compromise of passwords, financial details, journey history, or the names and birth dates of children holding Zip Oyster cards.

Enforcement ActionRansomware Group Operation

Ukraine and FBI Arrest Four Suspects in $40.5 Million Ransomware Investigation

Emerging Technology Risk · 8 sources · Updated 1h ago

Ukrainian authorities, working with the FBI, arrested four alleged key members of an international ransomware organization linked to more than $40.5 million in ransom payments from at least 28 companies worldwide. The investigation lasted more than two years and identified a group of at least 27 people connected to a broader network exceeding 60 members, including former Conti participants. Police conducted 66 searches across Kyiv and 11 Ukrainian regions, seizing electronic devices, bank cards and cash. The suspects face up to 15 years in prison if convicted; their individual affiliations with Conti, BlackSuit or Chaos remain unconfirmed.

Ai Platform SecurityInternet Facing Service Vulnerability

Meta Muse AI Accesses Private Messages as Shared Chat Links Expose Sensitive Data

Emerging Technology Risk · 5 sources · Updated 6h ago

Meta’s Muse AI agent accessed private messages through operating-system permissions, while a separate sharing flaw exposed thousands of conversation links without authentication. Reuters reviewed internal employee reports of apparent cross-account conversation confusion after Muse’s U.S. launch in September 2026. Testing with Signal and the AI Now Institute showed Muse could retrieve Signal messages through Android notification permissions granted during setup, without using Meta’s Connector tool. Requests on iPhone were blocked, but Muse retrieved messages on macOS by repurposing accessibility features, including screenshot tools. These findings concern access through device permissions, not a reported break of Signal’s encryption.

Privacy Surveillance PolicyAi Platform Security

Google AI Mode Adds Temporary Threads as Gemini Retains Temporary Chats for Two Months

Emerging Technology Risk · 2 sources · Updated 9h ago

Google is rolling out temporary threads in Search AI Mode that exclude conversations from visible history and future personalization, but it has not disclosed how long those conversations remain on its servers. Separately, Google’s Gemini Apps Privacy Hub now states that temporary chats and chats made with history off are retained for around two months, up from 72 hours. That Gemini retention policy is not confirmed to apply to AI Mode, whose temporary threads Google describes as detached from users’ accounts. Gemini documentation also describes human review and retention of reviewed conversations for up to three years; equivalent review arrangements for AI Mode remain unconfirmed.

Ai Platform Security

GAO Bypasses Safeguards on Four AI Platforms to Produce Viable Physical Attack Plans

Emerging Technology Risk · 3 sources · Updated 9h ago

The U.S. Government Accountability Office found that ChatGPT, Claude, Gemini and Grok generated information capable of supporting physical attacks during covert testing in spring 2026. All four platforms activated safety safeguards, but investigators bypassed them using common techniques and obtained the requested information in less than 12 hours per platform. Two platforms supplied information concerning improvised explosive devices, while two supplied information concerning attacks on the electrical grid. GAO validated the information’s viability through research, surveillance and expert interviews; the investigation involved controlled testing, not actual attacks.

Remote Access ImplantOffensive Tooling Release

ZRON Leak Exposes Previously Undocumented Tx Cross-Platform RAT

Malware Intelligence · 2 sources · Updated yesterday

A technical analysis of leaked ZRON materials attributes the previously undocumented Tx malware family, a modular, cross-platform remote access Trojan, to the company. The leaked source code, malware implants, software packages, product documentation, and internal discussions allowed researchers to reconstruct portions of ZRON’s offensive tooling. Researchers also examined publicly available malware samples and historical infrastructure records to trace the tooling beyond the leaked material.

Loader Delivery MechanismPayload Delivery Evasion

Fake Shipment ZIP Uses Kingsoft DLL Side-Loading in PureLogs-Linked Malware Chain

Malware Intelligence · 3 sources · Updated 2d ago

Attackers used a shipment-themed ZIP archive to launch a Windows infection chain combining a legitimate signed Kingsoft PDF application, DLL side-loading, and an obfuscated Python loader. A malicious DLL opened a decoy PDF while a disguised WinRAR utility extracted a password-protected package containing a Python runtime masquerading as Microsoft Edge Update. Successive XOR and Base64 decoding stages recovered a Windows executable that communicated with infrastructure associated with PureLogs and PureRAT campaigns; the findings do not establish that both malware families executed.

Loader Delivery MechanismSearch Ad Manipulation

FBI Warns Gootloader Compromised Over 1,000 WordPress Sites to Enable Ransomware

Malware Intelligence · 7 sources · Updated 2d ago

The FBI warned that Gootloader continues to enable ransomware intrusions through poisoned search results and compromised WordPress websites, with 2025–2026 campaigns primarily targeting U.S. legal professionals. Its November 16 FLASH alert identified more than 1,000 infected WordPress sites and 216 administrator-controlled domains as of October 2026, linking the activity to more than $28 million in ransomware payments. Victims searching for legal information encounter deceptive pages offering archives containing malicious JavaScript, leading to in-memory payload execution, credential theft, lateral movement and possible data exfiltration. The warning followed an Operation Endgame disruption, but the alert did not establish whether the listed domains overlapped with seized infrastructure.

Cybercrime Service EcosystemThreat Infrastructure Tracking

Researchers Link tcherber to FortiGate VPN Access Sales and Custom Malware Operation

Malware Intelligence · 5 sources · Updated 2d ago

Cyderes’ Howler Cell identified an active operation linked to the handle tcherber that combines enterprise VPN access brokerage, credential theft, custom malware development, and botnet infrastructure. Researchers recovered more than 500 FortiGate VPN credential entries and connection records showing successful root-level VPN tunnels into multiple victim environments during May and June 2026. Underground advertisements and shared contact information linked the operator to VPN access sales and malware services, supporting Cyderes’ assessment that the actor operates as both an initial access broker and a Malware-as-a-Service provider.

Open Source Dependency VulnerabilityWidely Deployed Product Advisory

Oracle Patches Java4Shell RCE in OpenJDK’s Accelerated Base64 Decoder

Patch & Detection Intelligence · 218 sources · Updated 45min ago

Oracle and OpenJDK released fixes for Java4Shell (CVE-2026-114938) after Calif disclosed a memory-corruption vulnerability in OpenJDK’s hardware-accelerated Base64 decoder that can enable unauthenticated remote code execution in exposed applications. Technical reports describe malformed input triggering an integer underflow and heap buffer overflow in an AVX-512 decoding path on supported x86-64 processors. Exposure depends on the exact JDK build, CPU capabilities, enabled JVM optimizations, and whether attacker-controlled input reaches the vulnerable decoder—not simply whether an application runs Java 17. Reports identify OpenJDK 17.0.6 and later as potentially affected, but precise vendor-specific scope requires checking advisories. The [OpenJDK advisory](https://openjdk.org/groups/vulnerability/advisories/2026-11-17) confirms the CVE was fixed and assigns it to hotspot/compiler with a CVSS 3.1 score of 8.1; researchers separately report a CVSS 4.0 score of 9.2.

Widely Deployed Product AdvisoryIdentity Authentication Vulnerability

Splunk Patches Critical SOAR Code Execution and Enterprise Authentication Bypass

Patch & Detection Intelligence · 4 sources · Updated 1h ago

Splunk released security updates addressing critical remote code execution vulnerabilities in SOAR and a separate authentication bypass in Enterprise. The most severe SOAR flaw, CVE-2026-131253, carries a CVSS score of 9.9 and allows an authenticated user with specific permissions to execute code as the SOAR service account. It affects versions before 8.8.0. SOAR 8.8.0 also fixes a critical REST API remote code execution vulnerability and a high-severity single sign-on bypass.

Endpoint Software VulnerabilityWidely Deployed Product Advisory

Chrome 158 Patches 204 Vulnerabilities, Including Six Critical Memory-Safety Flaws

Patch & Detection Intelligence · 12 sources · Updated 2h ago

Google released Chrome 158 for Windows, macOS, and Linux, fixing 204 security vulnerabilities: six critical, 62 high, 82 medium, and 54 low severity. The critical memory-safety flaws affect WebView, GPU, Media, garbage collection, ANGLE, and WebAudio and could potentially enable code execution within an affected browser process. Other fixes address security-restriction bypass, information disclosure, and resource-handling issues. Google's release notice does not report active exploitation or establish whether the critical flaws permit a sandbox escape.

Internet Facing Service VulnerabilityWidely Deployed Product Advisory

Ubiquiti Patches Seven UniFi Access Flaws, Including Root Code Execution

Patch & Detection Intelligence · 2 sources · Updated 5h ago

Ubiquiti disclosed seven vulnerabilities in UniFi Access Application, the platform used to manage electronic physical access controls. Five are rated critical and two high severity, affecting versions 4.3.22 and earlier. The flaws could enable root-level code execution on the application host, privilege escalation, information disclosure, and unauthorized changes or command execution on managed devices. Ubiquiti detailed the vulnerabilities in Security Advisory Bulletin 073, issued November 18, 2026.

Privacy Surveillance Policy

DeepMind Uses AI to Formally Verify Parts of Signal’s One-to-One Encryption

Regulatory & Policy · 7 sources · Updated 4h ago

Google DeepMind researchers used agentic AI to produce computer-checked proofs of selected cryptographic security properties in Signal’s libsignal library. The four-week proof of concept reportedly used 80 AI agents and the AlphaProof Code framework. It covered one-to-one session establishment and message-key updates in compiled AArch64 Linux code from libsignal v0.105.4. The results are not a security certification of the entire Signal application: group conversations, underlying cryptographic implementations and several other components were outside scope, and the proofs depend on stated assumptions about cryptographic primitives.

Cybersecurity RegulationEnforcement Action

US Launches Threat Disruption Center to Cut Off Scam Payments

Regulatory & Policy · 2 sources · Updated 7h ago

The US Homeland Security Task Force (HSTF) has launched a Threat Disruption Center to prevent victims’ money from reaching foreign scam, cybercrime, and sextortion networks. Based at the HSTF’s National Coordination Center in Fairfax, Virginia, the center combines investigative technology, technical support, and rapid intelligence sharing with private-sector partners. TRM Labs serves as the primary private-sector partner and built and maintains the center’s investigations platform.

Enforcement ActionState Sponsored Espionage

Russian National Charged in Alleged Surveillance of Palantir Employees

Regulatory & Policy · 4 sources · Updated 15h ago

U.S. authorities charged Russian national Igor Andreevich D’iu over an alleged Russian government-directed surveillance operation targeting employees of a Silicon Valley defense technology company understood to be Palantir Technologies. Prosecutors allege that handlers communicating through Telegram financed and directed him to photograph employees’ laptop screens, position remotely accessible laptop and Wi-Fi equipment near the company, and surveil employees and their homes using detailed dossiers. Investigators began examining D’iu after he allegedly followed a female employee for nearly 30 miles, prompting her to seek police assistance.

Trade Export ControlInformation Technology Sector Threat

Nvidia Creates Export-Control Oversight Committee Amid China Chip-Diversion Scrutiny

Regulatory & Policy · 3 sources · Updated yesterday

Nvidia established a National Security and Export Control Oversight Committee amid scrutiny of alleged efforts to divert restricted AI chips to China and Chinese companies’ remote access to overseas computing capacity. Chaired by Suzanne Nora Johnson, the board committee will meet quarterly, consult independent experts and oversee improvements to Nvidia’s security and trade-compliance teams. The move follows criticism from U.S. Treasury Secretary Scott Bessent over Nvidia’s safeguards. According to CNBC sources cited by Punto Informatico, Nvidia also warned customers that the proposed Remote Access Security Act could expose providers to legal action for giving Chinese companies remote access to AI-chip computing capacity.

Ransomware Group OperationInformation Technology Sector Threat

The Gentlemen Claims NVIDIA Breach Without Providing Evidence

Sector Threat Intelligence · 9 sources · Updated 4h ago

The Gentlemen ransomware group listed NVIDIA on its dark web leak site on November 18, 2026, alleging a compromise but providing no sample files, file inventory, claimed data volume or technical evidence. The listing referenced ransom negotiations and displayed a countdown pointing to the morning of November 28 UTC, which Ransomnews described as a publication deadline. NVIDIA had not publicly confirmed a breach, and no verified evidence established data theft or encryption of its systems. A pending-leak entry on Breach House documented the claim rather than independently validating it.

Internet Facing Service VulnerabilityEducation Sector Threat

UOC Virtual Campus Flaw Exposes Session-Related Identifiers in URLs

Sector Threat Intelligence · 2 sources · Updated 6h ago

INCIBE disclosed a medium-severity information exposure vulnerability in the Universitat Oberta de Catalunya (UOC) Virtual Campus, affecting its document-submission functionality and associated web forms. Tracked as CVE-2026-112216, the flaw has a CVSS v4.0 score of 5.3. After a document upload, the application places active-session-related identifiers, including apjsessionid, and internal request and user parameters in HTTP GET query strings. The published vector describes network-based exploitation requiring low privileges and no user interaction, with limited confidentiality impact.

Internet Facing Service VulnerabilityConsumer Discretionary Sector Threat

Hijacked Bitly Links Redirect Visitors From Old Brand and Trump Posts

Sector Threat Intelligence · 8 sources · Updated yesterday

Attackers hijacked expired Bitly links in genuine social media posts, redirecting visitors without editing the posts or compromising the accounts that published them. Links in McDonald’s and Burger King tweets from 2011 were redirected to unrelated Wikipedia articles, according to Ransomnews and Malwarebytes. Ransomnews also examined 97 Bitly links in Donald Trump’s archived tweets: 56 had creation dates later than the posts containing them, and seven of those re-created links led to unrelated destinations, including two domains resembling cryptocurrency services. One of the 97 links was blocked as malicious or spam.

Identity Impersonation FraudPhishing Campaign Intelligence

Fake Amazon Job Sites Collect Personal Data, Raise Parcel-Mule Fraud Concerns

Social Engineering & Fraud · 4 sources · Updated 2d ago

Multiple recruitment websites using Amazon-like names and branding are advertising purported remote assistant jobs paying up to $5,000 a month, despite not being operated by Amazon. Malwarebytes found that applications for the supposedly fully remote roles ask whether candidates have access to a vehicle and someone available to receive packages—questions that raise concerns about potential reshipping or parcel-mule fraud. The forms collect extensive personal information through Formspree, but the investigation could not establish who receives the submissions.

Business Email CompromiseConsumer Discretionary Sector Threat

Phishing Compromises Luxury Hotel Reservation Inboxes to Enable Refund Fraud

Social Engineering & Fraud · 2 sources · Updated 3d ago

Attackers used fake booking requests to compromise luxury hotel reservation inboxes, then intercepted legitimate refund requests and substituted fraudulent versions that led to unauthorized financial approvals. SC Media reported the scheme, citing an account attributed to Schneier on Security. The attack exploited a routine business workflow: reservation staff must respond to inquiries from unfamiliar senders, making blanket advice to avoid suspicious messages difficult to apply.

Phishing Campaign IntelligenceIdentity Impersonation Fraud

Attacker Abuses LearnWorlds Accounts to Send 13,000 YouTube OAuth Phishing Emails

Social Engineering & Fraud · 8 sources · Updated 3d ago

An attacker abused LearnWorlds’ account-registration and welcome-email features to send approximately 13,000 phishing emails between November 7 and 14, using around 570 free trial accounts and one paid account. The messages originated from LearnWorlds infrastructure and targeted externally supplied recipient lists with fraudulent YouTube copyright-infringement notices. Recipients were directed through a YouTube redirect to a YouTube-branded phishing site presenting a genuine Google OAuth authorization flow, intended to trick users into granting access to their Google and YouTube accounts. The number of recipients who clicked or authorized access has not been established.

Phishing Campaign IntelligenceIdentity Impersonation Fraud

Fake FC 27 Prime Gaming Rewards Site Steals EA Passwords and 2FA Codes

Social Engineering & Fraud · 2 sources · Updated 3d ago

A phishing operation is impersonating an “FC27 x Prime Gaming” reward offer to steal EA account email addresses, passwords, and two-factor authentication (2FA) codes, Malwarebytes reported. Victims encounter a copied EA sign-in page on attacker-controlled infrastructure that sends their information to fedserv.thenewsip[.]site. Server responses determine whether the page displays a password error or requests a verification code, but the captured client-side code does not establish how the backend validates credentials.

Cybercrime Service EcosystemIndustrials Sector Threat

Nine Underground Listings Advertise Enterprise VPN, Cloud and CI/CD Access

Supply Chain Risk · 2 sources · Updated 3d ago

ThreatMon documented nine underground advertisements offering alleged access to enterprise VPNs, cloud accounts, software delivery infrastructure and business applications. Posted between November 2 and November 9, 2026, the offers ranged from individual remote-access accounts to bundles of cloud keys, database credentials, API tokens and authenticated sessions. One seller claimed privileged TeamCity access to a Nordic warehouse-automation group’s build and deployment environment spanning 130 customer environments. Other listings targeted manufacturers, an ERP SaaS business and a game publisher; an authenticated PandaDoc session allegedly exposed insurance records and personal information.

Vendor Distribution CompromiseGovernment Diplomatic Threat

Malicious QBase Update Prompts German Military Drone Grounding

Supply Chain Risk · 16 sources · Updated 5d ago

A cyberattack on German drone manufacturer Quantum Systems reportedly compromised its software-update distribution server and delivered a malicious update for QBase, its ground-control and mission-planning software. Reporting by Der Spiegel and the Financial Times indicated that the update reached German military servers and other customers, prompting the Bundeswehr to ground Quantum Systems drones as a precaution. The drones themselves were reportedly not directly infected, and the number of customers who installed the update remains unclear.

Remote Access ImplantPayload Delivery Evasion

Wpcdoor Backdoor Delivered Through Compromised Korean DVR Management Software

Supply Chain Risk · 2 sources · Updated 7d ago

AhnLab ASEC reported confirmed Wpcdoor backdoor infections in South Korea involving compromised DVR Central Management Software (CMS) developed by an unnamed Korean CCTV/DVR company. The campaign uses two stages of DLL side-loading to download and execute Wpcdoor in memory. Researchers also identified an alternative infection chain using different DLLs and modified environment variables. AhnLab attributes the activity to Larva-26013, which it reports has also operated outside South Korea.

Ai Platform SecurityPackage Repository Poisoning

Poisoned Models and Reclaimed Namespaces Expose AI Coding-Agent Supply-Chain Risks

Supply Chain Risk · 4 sources · Updated 7d ago

Microsoft Defender Research and the Microsoft AI Red Team demonstrated how poisoned fine-tuning data and attacker-trained model checkpoints can backdoor coding agents into executing malicious tool calls. Poisoning approximately 1% of training examples produced roughly 78% attack success in familiar agent harnesses, though transfer to an unseen harness was limited. Backdoored Qwen3.5 checkpoints achieved 84.1%–94.8% success on unseen repositories and transferred across multiple harnesses. Subsequent reinforcement learning recovered much of their legitimate coding performance without substantially weakening the backdoors. The Kerkoporta research paper reports structured-output backdoors with success rates up to 96.2%, showing how a trigger in an agent’s context can cause an attacker-chosen tool call. These controlled experiments demonstrate that strong coding benchmarks do not establish model integrity.

Open Source Dependency VulnerabilityInternet Facing Service Vulnerability

Hickory DNSSEC Flaw Allows Forged Zone-Apex NODATA Responses to Be Accepted as Secure

Vulnerability Disclosure · 3 sources · Updated 2d ago

A DNSSEC validation vulnerability in Hickory, tracked as CVE-2026-123600, allows forged NODATA responses at a DNS zone apex to be accepted as secure. An attacker able to answer or modify DNS responses can replay genuine signed NSEC3 and SOA records from the zone without providing an NSEC3 record matching the hashed query name. A recursive resolver may then cache the false denial of record existence for the negative TTL, suppressing legitimate DNS answers. The affected validation branch treated a query-name match with the SOA name as sufficient proof despite the missing NSEC3 match.

Financial Sector ThreatInternet Facing Service Vulnerability

Hyderabad NBFC Loses Over ₹3.3 Crore in 87-Minute Attack on Systems and APIs

Vulnerability Disclosure · 3 sources · Updated 2d ago

A Hyderabad-based non-banking financial company (NBFC) allegedly lost more than ₹3.3 crore through 183 unauthorized transfers in 87 minutes after attackers accessed its systems and APIs on October 19. According to its chief financial officer’s complaint, the money went to 52 suspected beneficiary accounts. The company detected the breach that day and reported it through India’s National Cyber Crime Reporting Portal; the Telangana Cyber Security Bureau registered a case following the October 20 complaint. Police reportedly blocked ₹40 lakh in beneficiary accounts, with the investigation continuing. ([Siasat](https://www.siasat.com/hyderabad-nbfc-loses-rs-3-3-crore-in-87-minute-cyberattack-3562826))

Government Diplomatic ThreatMilitary Defense Sector Threat

NATO Expands AI Cyber Defense With OpenAI Models as European Agencies Find Vulnerabilities

Vulnerability Disclosure · 3 sources · Updated 2d ago

NATO’s Communications and Information Agency is providing NATO’s core institutions with access to OpenAI cybersecurity models through Daybreak to help defend Allied networks. Help Net Security reported that NATO’s AI cyber defense work uses up to 300 billion OpenAI tokens per month. Ukraine is also using Daybreak to protect civilian infrastructure, while participants in NATO’s 2026 Crossed Swords exercise built and tested AI agents with an emphasis on human control.

Open Source Dependency VulnerabilityEndpoint Software Vulnerability

macOS Unicode Path Flaws Bypass Security Checks in Four Developer Tools

Vulnerability Disclosure · 3 sources · Updated 2d ago

Four vulnerabilities affecting libarchive, Deno, HashiCorp go-slug, and Python setuptools can allow local attackers to bypass file exclusions or permission checks on macOS. The flaws arise when tools compare filenames as raw bytes while the filesystem treats different Unicode representations or spellings, such as ß and ss, as equivalent. Depending on the affected tool and workflow, exploitation could overwrite protected files, include sensitive files in uploaded or published artifacts, or enable code execution. CERT/CC reports no exploitation in the wild, and the attacks require local access.

Find out if you're exposed

Don't read about it. Know when it affects you.

Mallory correlates every story on this page with your attack surface (assets, vendors, identities, subsidiaries) and surfaces a small set of evidence-based cases instead of 10,000 alerts.

Or grab the RSS feeds →