Weekly Threat Briefing — 2026-10-26 to 2026-11-09
99 findings390 articles6 sections
- Critical findings
- 0
- High findings
- 0
- CISA KEV additions
- 0
- First seen in this period
- 390 Articles3,276 CVEs62,492 IOCs
First seen in this period: news articles, each counted once · CVEs rated CVSS 7.0+ (NVD, GitHub, WordPress) · CISA KEV catalog additions · new abuse.ch indicators (ThreatFox, URLhaus, Feodo Tracker), dated by abuse.ch.
Priority queue
Act on these first
Critical and high findings from this briefing and its live updates. Most severe first, then newest.
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: KEV
- Critical severitySection VulnerabilitiesSource: RSS
- Critical severitySection VulnerabilitiesSource: CVE
- Critical severitySection VulnerabilitiesSource: CVE
- Critical severitySection VulnerabilitiesSource: CVE
Executive summary · as of 07:05 UTC
The situation
This window was dominated by actively exploited edge and web-stack vulnerabilities and the operational use of agentic AI by attackers. CISA added 13 CVEs to KEV. They include a SonicWall SSL-VPN SAML/MFA bypass (CVE-2026-19868), a chainable Ivanti Connect Secure trio, two unauthenticated PHP RCEs, a WordPress Core upgrade race condition, GitLab CE/EE RCE and Adobe Commerce arbitrary upload. The older Apple CVE-2024-27834 was also newly added. QNAP warned of in-the-wild Plex2Shell exploitation, and a decade-old OpenSSH-on-musl authentication flaw puts Alpine-based containers, routers and IoT devices at critical risk. NVD published 3,237 CVEs in the window, of which 60 critical-rated entries were reviewed, alongside 312 OSV advisories and 649 WordPress component vulnerabilities.
On the threat-actor side, Microsoft and Mandiant documented abuse of the agentic offensive framework ARTEX. Storm-3358 used it against the financial sector, and other actors used it to harvest OAuth secrets from GitHub. Mandiant also detailed LLMjacking intrusions. Coordinated attacks on European supermarket refrigeration systems on November 2-3 prompted a joint FBI/CISA advisory. Russian state actors ran APT29 mobile watering holes and a zero-click smartphone espionage campaign. Google and partners dismantled the 1.3-million-device XCPE set-top-box proxy botnet. Crédit Mutuel Arkéa suffered a $133 million fraud via a service provider.
Credential exposure is the leading breach theme. Chainalysis confirmed credentials and live tokens for all customers were exposed. mSecure forced password resets, and Applebee's leaked an embedded app password. N0n claims an 830-million-row Expedia loyalty dataset. Other incidents include Iris Ohyama (7.7M customers), Revolve (1.5M cards), DigiCert, and Amadeus-linked airline disclosures. abuse.ch logged 62,372 new indicators across 776 families, led by GitHub-hosted SalatStealer/zapret lures, AMOS, IClickFix/ClearFake and Vidar. 557 ransomware and leak-site postings were recorded, with LockBit5, Qilin, Deadlock and N0n most active in the sample.
Browse the full briefing
99 findings · 6 sections- 19 critical
- 46 high
- 24 medium
- 6 low
- 4 info
- Vulnerabilities31 findingsActively Exploited & Critical Vulnerabilities17 critical · 13 high · 1 medium
- Threat Actors13 findingsThreat Actors, Campaigns & Disruptions8 high · 4 medium · 1 low
- Malware16 findingsMalware, Phishing & Supply-Chain Campaigns10 high · 6 medium
- Data Breaches & Leaks19 findingsData Breaches & Leaks2 critical · 11 high · 5 medium · 1 low
- Tools & Techniques10 findingsResearch, Tradecraft & Defensive Tooling3 high · 4 medium · 3 info
- Policy10 findingsPolicy, Regulation & Law Enforcement1 high · 4 medium · 4 low · 1 info
Actively Exploited & Critical Vulnerabilities
CISA added 13 CVEs to KEV, concentrated on remote-access gateways (SonicWall, Ivanti), web runtimes and platforms (PHP, WordPress Core, GitLab, Adobe Commerce), and infrastructure (Cisco ASA/FTD, VMware ESX, Linux kernel). Vendor reporting adds exploited QNAP Plex2Shell, a critical OpenSSH-on-musl auth bypass and unpatched D-Link switch zero-days. Beyond the 60 NVD, 40 OSV and 40 WordPress entries reviewed, a further 3,177 NVD CVEs, 272 OSV advisories and 609 WordPress vulnerabilities were published in the window.
- Critical severitySonicWall SonicOS SSL-VPN MFA bypass exploited (CVE-2026-19868) (opens the source in a new tab)
Added to CISA KEV on 2026-11-06. A user holding valid credentials for an alternate authentication method can bypass SAML authentication on SonicOS SSL-VPN, including SAML-enforced MFA. SonicWall's update fixes this and six other SonicOS flaws (per BleepingComputer). Patch immediately, review SSL-VPN authentication logs for logins outside the SAML flow, and disable unused local or LDAP auth methods.
- Critical severityIvanti Connect Secure exploit chain added to KEV (CVE-2026-111357, CVE-2026-118527) (opens the source in a new tab)
Two Ivanti Connect Secure CVEs were added to KEV on 2026-10-30. CVE-2026-118527 uses predictable identifiers to let an authenticated attacker upload attacker-controlled files to a known disk location. CVE-2026-111357 is missing authentication on a critical function, enabling privilege escalation. Both chain with CVE-2026-104935. Patch, run Ivanti integrity checks and hunt for unexpected files on appliances.
- Critical severityPHP unauthenticated RCE pair exploited (CVE-2026-109619, CVE-2026-109618) (opens the source in a new tab)
Two PHP flaws were added to KEV on 2026-10-26, both allowing unauthenticated remote code execution. CVE-2026-109619 is a stale internal request-state flaw; CVE-2026-109618 is memory corruption. The impact spans internet-facing PHP applications broadly. Prioritise runtime upgrades across web fleets and container base images.
Source: KEV - Critical severityWordPress Core upgrade TOCTOU race condition exploited (CVE-2026-119977) (opens the source in a new tab)
Added to KEV on 2026-11-03. Full-package core upgrades use predictable, web-accessible staging paths. An unauthenticated attacker can execute code on the server during the upgrade window. Review web logs for requests to upgrade staging directories coinciding with core updates.
Source: KEV - Critical severityGitLab CE/EE unauthenticated RCE via CI/CD regex (CVE-2026-97229) (opens the source in a new tab)
Added to KEV on 2026-10-29. An integer overflow lets an unauthenticated user execute arbitrary code on the GitLab server via a crafted regular expression in CI/CD configuration. Upgrade self-managed instances and audit recent pipeline configuration changes.
Source: KEV - Critical severityAdobe Commerce/Magento arbitrary file upload exploited (CVE-2026-48356) (opens the source in a new tab)
Added to KEV on 2026-10-27. Unrestricted upload of dangerous file types can lead to arbitrary code execution. E-commerce operators should patch and sweep media and upload directories for web shells and skimmer injections.
- Critical severityOpenSSH on musl: long-password authentication bypass hits Alpine, routers, IoT (opens the source in a new tab)
A decade-old flaw affects OpenSSH running on the musl C library, including Alpine and other musl-based distributions. Any password between 257 and 1,024 characters reportedly authenticates. Containers, routers and IoT devices are exposed. Emergency updates are available: rebuild Alpine-based images and disable password authentication where possible.
- Critical severityMicrosoft Entra Device Registration authorization bypass (CVE-2026-109469) (opens the source in a new tab)
CVSS 10. An authorization bypass through a user-controlled key in Entra Device Registration lets an unauthorized network attacker elevate privileges. Monitor for anomalous device registrations and join events.
Source: CVE - Critical severityOpenAM unauthenticated session-token disclosure (CVE-2026-123551) (opens the source in a new tab)
CVSS 10. OpenAM 14.0.0 to before 16.1.4 lets unauthenticated attackers read live session tokens from the notification WebSocket. Attackers bypass the agent-only check via legacy Liberty ID-FF endpoints. Upgrade to 16.1.4 and invalidate sessions.
Source: CVE - Critical severityCanonical LXD authentication bypass and image-based host escapes (CVE-2026-112816) (opens the source in a new tab)
CVSS 10. Improper certificate validation in the LXD HTTPS listener lets an unauthenticated attacker authenticate as any trusted client, gaining full cluster and VM access. Fixed in 4.0.15, 5.0.11, 5.21.9 and 6.10.1. Six further critical GitHub advisories enable host file read/write and command execution via malicious images, symlinks and backup-compression argument injection (e.g. GHSA-47w9-6r3f-938g, GHSA-vghh-5rfx-xhq8).
- Critical severityIBM API Connect: multiple authenticated RCE and cross-tenant flaws (opens the source in a new tab)
Six CVSS 9.9 flaws affect IBM API Connect 10.0.8.0-10.0.8.12 and 12.1.0.0-12.1.1.2. They include OS command injection (CVE-2026-17402), code injection (CVE-2026-17359), unvalidated upload RCE (CVE-2026-17399), cross-tenant LDAP injection (CVE-2026-17392), tenant data access (CVE-2026-17410) and path traversal file write (CVE-2026-17354). Multi-tenant deployments are most at risk.
- Critical severityProgress WS_FTP Server unauthenticated credential disclosure (CVE-2026-84295) (opens the source in a new tab)
CVSS 10. Improper authentication in WS_FTP Server on Windows before 2024.0.6 and 2025.0.3 lets unauthenticated attackers obtain sensitive data, including administrative credentials. Managed file transfer platforms are historically prime targets for extortion crews.
- Critical severityApple WebContent sandbox escapes (CVE-2026-112690, CVE-2026-113897) (opens the source in a new tab)
CVSS 10. Two flaws let a remote attacker break out of the WebContent sandbox. Fixed in iOS/iPadOS 27.2 and 26.7.2, macOS Golden Gate 27.2, Sequoia 15.8.2, Tahoe 26.7.2, visionOS 27.2 and Safari 27.2. Prioritise mobile fleets given concurrent Russian mobile espionage reporting.
Source: CVE - Critical severityAdobe Campaign Classic SQL injection to code execution (CVE-2026-105781) (opens the source in a new tab)
CVSS 10. A SQL injection in Adobe Campaign Classic can result in arbitrary code execution in the context of the current user.
Source: CVE - Critical severityMLflow unauthenticated RCE via scorer deserialization (CVE-2026-116692) (opens the source in a new tab)
GHSA-26p8-2jq9-3vq9, CVSS 9.8. Unauthenticated MLflow servers allow remote code execution via third-party scorer deserialization that imports caller-placed artifact modules. MLflow servers are frequently exposed without authentication.
Source: CVE - Critical severityRust zip crate parser differential enables file overwrite/RCE (CVE-2026-115905) (opens the source in a new tab)
GHSA-m665-v7pm-7grj, CVSS 10. The Rust zip crate through 9.0.4 has inconsistent Central Directory and Local File Header filename handling. This 'zip cloaking' enables security-control bypass, arbitrary file overwrite and RCE.
Source: CVE - Critical severityWordPress WP Libre Form unauthenticated PHP upload RCE (CVE-2026-118423) (opens the source in a new tab)
CVSS 10. WP Libre Form 1.5.0.1 does not validate uploaded file types, so unauthenticated attackers can upload and execute PHP files. AjaxLoadMore Cache <= 3.1.1 has a separate unauthenticated RCE (CVE-2026-61991).
- High severityGoogle Brotli decoder memory corruption (CVE-2026-111497) (opens the source in a new tab)
Added to KEV on 2026-10-27. An unchecked return value lets a crafted Brotli-compressed stream corrupt decoder state memory. Outcomes include memory disclosure, denial of service or possible code execution. Identify bundled libbrotli copies in applications and appliances.
Source: KEV - High severityApple pointer-authentication bypass newly added to KEV (CVE-2024-27834) (opens the source in a new tab)
This 2024 CVE was newly added to CISA KEV on 2026-11-04. An attacker with arbitrary read/write capability can bypass pointer authentication on iOS, iPadOS, macOS, Safari, tvOS and watchOS. It is typically a stage in exploit chains. Ensure legacy Apple devices are updated or retired.
Source: KEV - High severityLinux kernel ESP page-cache modification flaw exploited (CVE-2026-46300) (opens the source in a new tab)
Added to KEV on 2026-10-28. During socket buffer coalescing, the shared-fragment marker is not preserved, so ESP input processing decrypts shared fragments in place. This allows unintended modification of page-cache-backed data. Apply distribution kernel updates.
Source: KEV - High severityCisco ASA/FTD unauthenticated DoS exploited (CVE-2026-76418) (opens the source in a new tab)
Added to KEV on 2026-10-29. A NULL pointer dereference lets an unauthenticated remote attacker force device reloads. The resulting perimeter DoS can mask concurrent intrusion activity.
Source: KEV - High severityBroadcom VMware ESX insufficient logging exploited (CVE-2026-41709) (opens the source in a new tab)
Added to KEV on 2026-10-28. An actor with admin access can perform certain operations without generating logs. This is a significant blind spot for ransomware investigations on hypervisors. Patch, and forward ESX logs off-host.
Source: KEV - High severityQNAP NAS devices breached via Plex2Shell chain (opens the source in a new tab)
QNAP reports active exploitation of the Plex2Shell vulnerability chain against Plex Media Server on its NAS devices. Update Plex via QNAP App Center and remove internet exposure of NAS management and media services.
Source: RSS - High severityD-Link DGS-3130 switch zero-days, no patch yet (opens the source in a new tab)
D-Link disclosed multiple unpatched zero-days in DGS-3130 series managed stackable switches and published mitigations only. Restrict management-plane access to trusted networks.
Source: RSS - High severityCritical Rails Active Storage/ImageMagick file read-write flaw (opens the source in a new tab)
A critical bug in the Rails Active Storage ImageMagick integration lets crafted image uploads write arbitrary files and read sensitive files. This could enable lateral movement. Update Rails and restrict ImageMagick policies.
Source: RSS - High severitySOHO router exploit flood: Linksys, WAVLINK, TOTOLINK, Qihoo 360, Tenda (opens the source in a new tab)
Dozens of CVSS 9.9-10 flaws with public exploits were published for consumer and SOHO devices. They comprise OS command injection, stack overflows, hard-coded credentials and auth bypass. Affected vendors are Linksys (E5350, E5600, E8450), WAVLINK, TOTOLINK, Qihoo 360, Tenda AC19, Edimax BR-6208AC, Comfast and D-Link DIR-505. These devices are likely Mirai/Mozi recruitment targets.
- High severityApache Juneau: batch of critical REST server flaws (opens the source in a new tab)
CVSS 10 flaws were published across Apache Juneau. They include class-loading code execution (CVE-2026-73098), credential header leakage on redirects, unredacted Authorization headers in logs, CRLF response splitting, CSRF via the content parameter, and DirectoryResource path traversal (CVE-2026-121489). Update Juneau-based REST services.
Source: CVE - High severitykaniko leaks Entra ID tokens to lookalike ACR hosts (CVE-2026-105769) (opens the source in a new tab)
GHSA-5q7h-w9jf-cx97, CVSS 9.3. The chainguard-dev kaniko fork can send a build's Azure Entra ID access token to lookalike ACR registry hosts. This risks cloud token theft from CI pipelines.
Source: CVE - High severityAtlantis command injection via Terraform workspace (CVE-2026-116676) and AI-framework RCEs (opens the source in a new tab)
Atlantis allows OS command injection through PR-controlled Terraform workspace names, giving RCE during plan (GHSA-rjh2-4xgw-p2hj). It sits within a wider set of AI/automation framework flaws: Agno eval() code execution (CVE-2026-35002), NVIDIA NeMo Agent Toolkit missing authentication (CVE-2026-65145) and a Prefab UI sandbox escape (CVE-2026-120338).
Source: CVE - High severityWordPress plugin critical-vulnerability wave (opens the source in a new tab)
649 WordPress component vulnerabilities were recorded in the window; the 40 reviewed were all CVSS 9.1+. They include CVE-2026-113617 (9.9), Gravity Forms (CVE-2026-103039), s2Member (CVE-2026-86596), JupiterX Core social login (CVE-2026-106013), Uncanny LearnDash Toolkit (CVE-2026-119096) and WP Project Manager (CVE-2026-116512). Inventory plugins and enable auto-updates.
Source: CVE - Medium severityChrome 157 fixes 213 security issues, including 7 critical (opens the source in a new tab)
Chrome 157 addresses 213 security issues, 7 of them critical. Verify enterprise browser versions rather than relying on auto-update.
Source: RSS
Threat Actors, Campaigns & Disruptions
Agentic AI moved from proof of concept to operational use, with ARTEX abused for OAuth secret harvesting and financial-sector targeting, alongside LLMjacking. Russian-aligned activity spanned cold-chain sabotage in Europe, APT29 mobile watering holes and zero-click smartphone espionage. A major industry takedown dismantled the XCPE proxy botnet.
- High severityStorm-3358 weaponizes ARTEX against financial sector (opens the source in a new tab)
Microsoft reports that financially motivated Storm-3358 began targeting financial-sector organizations in October 2026 using the agentic AI offensive framework ARTEX. This expands abuse of the tool by cybercriminals.
Source: RSS - High severityARTEX agents harvest OAuth app secrets from GitHub (opens the source in a new tab)
Microsoft observed actors running ARTEX agents to search GitHub for OAuth application secrets. They then compromise the associated apps and perform authenticated actions in the owning tenant. Rotate exposed app secrets, enforce secret scanning and monitor service principal sign-ins.
- High severityCoordinated attacks on European supermarket refrigeration systems (opens the source in a new tab)
On November 2-3, store fridges and freezers warmed overnight in six European countries after attacks on refrigeration and cold-chain control systems. Russian-aligned groups are credible suspects, but attribution remains open. Unit 42 and Dark Reading also covered the attacks, which cascaded across shared vendor and remote-access dependencies.
- High severityAPT29 watering holes target thousands of mobile users (opens the source in a new tab)
APT29 compromised websites to fingerprint mobile visitors, then harvested information from selected targets. Thousands of mobile users were exposed.
Source: RSS - High severityRussian zero-click espionage against iPhone and Android (opens the source in a new tab)
Russian state-linked hackers ran a zero-click campaign to steal sensitive data from iPhone and Android devices of users in Ukraine, the US and Europe. Keep mobile OS patching current, given the WebContent sandbox escapes fixed this window.
Source: RSS - High severityGoogle and partners dismantle 1.3M-device XCPE set-top-box botnet (opens the source in a new tab)
Google and industry partners dismantled the XCPE backdoor botnet of over 1.3 million Android TV set-top boxes. The boxes shipped with malware that rented out users' connections as residential proxies and conducted ad fraud. Google assesses with high confidence the operation was run for profit by manufacturer Videostrong.
- High severity$133M Crédit Mutuel Arkéa heist via service provider (opens the source in a new tab)
French cooperative bank Crédit Mutuel Arkéa was attacked through a service provider, enabling up to $133 million in fraudulent transactions. The bank suspended certain systems in response.
Source: RSS - High severityMandiant: LLMjacking monetizes compromised cloud and model access (opens the source in a new tab)
Mandiant responded to multiple 2026 incidents in which actors hijacked enterprise LLM access ('LLMjacking') via compromised cloud credentials and monetized model resources. Monitor model-invocation spend and API key usage.
- Medium severityMandiant: ARTEX and other agentic offensive frameworks (opens the source in a new tab)
Mandiant analyses recent highly automated intrusions using ARTEX and similar agentic offensive frameworks, with detection and hardening guidance. Separately, researchers found hundreds of publicly exposed copies of the Chinese-made Artex tool online (DataBreaches.net/TechCrunch).
Source: RSS - Medium severityNorth Korean actors expand AI use in tradecraft (opens the source in a new tab)
Microsoft details varying levels of AI adoption across North Korean threat actors to speed up and scale their operations.
Source: RSS - Medium severityWhite House devices running rentable residential proxies (opens the source in a new tab)
Krebs reports that several White House staffers' devices run residential proxy services, making their IP addresses rentable by third parties. One provider is Russian.
Source: RSS - Medium severityRansomware disrupts Queens Public Library (opens the source in a new tab)
Queens Public Library confirmed a ransomware attack disrupted Wi-Fi and on-site computers across its 66-branch system.
Source: RSS - Low severityRussian AI-assisted influence operation targets Latin America (opens the source in a new tab)
An AI-assisted Russian influence operation used fake emails, forged government documents and a fabricated local organization to target schools in Peru and Ecuador, among other activity.
Source: RSS
Malware, Phishing & Supply-Chain Campaigns
abuse.ch recorded 62,372 new indicators (30,029 ThreatFox, 32,343 URLhaus) across 776 families; 45,198 were included in the supplied data. GitHub-hosted lures dominate: fake zapret/Telegram-proxy tools delivering SalatStealer and FakeGit repositories pushing SmartLoader. AMOS, IClickFix/ClearFake and Vidar remain high-volume. Developer and researcher tooling is a recurring delivery vector (fake security tools, a malicious IDE extension, a game launcher compromise), and device-code and session-theft phishing continues against Microsoft 365.
- High severitySalatStealer via fake zapret/Telegram-proxy tools on GitHub
URLhaus logged roughly 20,000 malware-download URLs across overlapping tags (github/SalatStealer/zapret 9,267; github/zapret 4,977; github/SalatStealer 2,947; exe/github/SalatStealer 2,224). These are fake zapret, Discord/YouTube unblock and Telegram proxy tools on GitHub releases delivering SalatStealer. ThreatFox added a further 1,110 SalatStealer hashes and C2 entries. Samples: hxxps://github[.]com/TraderAlembic92/tg-ws-proxy/releases/download/Tg-Ws-Proxy/TgWsProxy_windows_arm64[.]exe, hxxps://github[.]com/GalaxyHornet/tgwsproxy/releases/download/tgwsproxy/TgWsProxy_windows[.]exe, hxxps://github[.]com/SymbolTeller/tg-ws-proxy-v2/releases/download/Tg-Ws-Proxy-v2/TgWsProxy_windows_arm64[.]exe.
- High severityAMOS macOS stealer: 3,523 new indicators
3,523 ThreatFox indicators (sha256, URL, domain, ip:port) cover payloads, payload delivery and botnet C2 for the macOS infostealer AMOS. Samples: 767602240f255c7c08ede0cc694d7f25944c57f7ad57dad4ec69af4e3173ddf4, ef85166c806e2685a5a0e56d77c574ef21af6adbeb9d8067ac82de287897a960, 054e9f8e29c25fa2e0715932e581ca1f1797376c4cc8526f933ec4a84b264445.
Source: abuse.ch - High severityIClickFix social-engineering infrastructure: 2,987 indicators
2,987 ThreatFox indicators (domain, URL, ip:port) cover payload delivery and C2 for the IClickFix fake-CAPTCHA paste-and-run lure. Samples: fsroma[.]com, codes[.]caffe-img[.]com, slotcore[.]top.
Source: abuse.ch - High severityClearFake fake-update injections: 2,305 indicators
2,305 ThreatFox indicators (domain, URL, sha256) cover the ClearFake compromised-site injection framework, including payloads staged on the jsDelivr CDN. Samples: www[.]vinte3comunicacao[.]com[.]br, hxxps://fastly[.]jsdelivr[.]net/gh/sdf4563jvn/b7c13ff9c733/b77d-16f9f9d0817c, betballl90[.]app.
- High severityVidar infostealer: 2,390 indicators
2,390 indicators across ThreatFox and URLhaus (URLs, domains, ip:port, hashes) cover Vidar C2, payloads and downloads. Samples: hxxps://at[.]adwgamestore[.]org/, at[.]adwgamestore[.]org, hxxps://c[.]aot88[.]com/.
Source: abuse.ch - High severityFake security tools on GitHub drop backdoors and infostealers (opens the source in a new tab)
Trojanized security projects on GitHub target cybersecurity professionals, deploying a backdoor and an information stealer. Vet researcher tooling and run untrusted PoCs only in isolated sandboxes.
- High severityNIKKE PC launcher supply-chain compromise (opens the source in a new tab)
Attackers tampered with files used by the Windows launcher for Goddess of Victory: NIKKE, putting players' stored passwords and personal files at risk. Affected users should scan systems and rotate credentials.
- High severityMalicious Claude Code-branded extension plants persistent SSH backdoor (opens the source in a new tab)
Microsoft analysed a malicious Claude Code-branded Cursor extension that installs a persistent SSH backdoor on developer machines. Microsoft provides detection and remediation guidance. Audit IDE extensions and authorized_keys files.
- High severitySkyW4lker device-code phishing kit targets Microsoft 365 (opens the source in a new tab)
SkyW4lker (x365 Sentinel) is an active device-code phishing kit targeting Microsoft 365 users in the US, Europe and elsewhere. It abuses the legitimate OAuth device authorization flow to obtain tokens without stealing passwords. Restrict device-code flow via Conditional Access.
- High severityFake OpenAI/SpaceX event invites steal Microsoft and Google sessions (opens the source in a new tab)
Fake OpenAI, SpaceX and Anthropic event invitations lure employees to two phishing-as-a-service platforms. The platforms steal Microsoft and Google credentials and authentication sessions.
- Medium severitySmartLoader via FakeGit repositories: ~2,850 URLs
URLhaus logged 2,082 (github/SmartLoader/zip) plus 771 (FakeGit/SmartLoader) malware-download URLs. They are ZIP archives in fake GitHub projects such as game mods and dev tools delivering SmartLoader. Samples: hxxps://github[.]com/mnochtioui/redux_v1/raw/refs/heads/main/src/Redux1/v-redux-2[.]2-beta[.]5[.]zip, hxxps://github[.]com/Misaya0/PSD-OWOD/raw/scv/SAM_H_json/PS-OWOD-1[.]2[.]zip, hxxps://github[.]com/Michelinboard436/inference-learn/raw/main/[.]github/learn_inference_v3[.]9[.]zip.
- Medium severityMirai and Mozi IoT botnets: ~3,700 indicators
Mirai accounted for 2,184 indicators (hashes, URLs, C2 ip:port, domains) and Mozi for 1,053 plus 497 MIPS ELF download URLs, consistent with mass router exploitation. Samples: 26c92d1fbffdfdb66755e5a56f81d0c7dd7d0ef4698bcc486d157fc9684ba377, hxxp://112[.]249[.]89[.]169:40029/i, hxxp://42[.]235[.]141[.]158:37542/bin[.]sh.
- Medium severityCommodity C2: VShell, AsyncRAT, Cobalt Strike, PureRAT
New indicators cover VShell (718), AsyncRAT (695), Cobalt Strike (424) and PureRAT (304), mostly C2 ip:port, hashes and domains. Samples: 106[.]75[.]148[.]37:9999 (VShell), 110[.]41[.]77[.]122:443 (Cobalt Strike), 154[.]29[.]79[.]11:56001 (PureRAT). Block and hunt for these in egress logs.
Source: abuse.ch - Medium severityStrelaStealer and WARDENStealer infrastructure
ThreatFox added 490 StrelaStealer delivery and C2 domains/URLs and 365 WARDENStealer indicators. Samples: church[.]springsoflife[.]org (StrelaStealer), hxxps://waka-tune[.]mom/ (WARDENStealer), hxxps://dibeert[.]pics/ (WARDENStealer).
Source: abuse.ch - Medium severityTrojanized Inno installers in fake-download campaign (opens the source in a new tab)
Microsoft tracks a fake-download campaign using trojanized Inno Setup installers to push multiple monetization payloads, with guidance on investigating the full installer session.
- Medium severityMicrosoft Q3 2026 email threat landscape (opens the source in a new tab)
Email phishing rose sharply at the end of Q3 2026. QR-code phishing declined while CAPTCHA-gated phishing recovered, and spoofing of familiar business-service sender domains increased.
Source: RSS
Data Breaches & Leaks
Credential and token exposure leads this window. Chainalysis confirmed exposure of credentials and live tokens for all customers, mSecure forced password resets, and Applebee's leaked an embedded app password. The largest claimed dataset is N0n's 830M-row Expedia loyalty dump, followed by Iris Ohyama (7.7M) and Revolve (1.5M cards). Third-party compromises (Amadeus, Iris Ohyama's cloud provider) remain a key vector. Leak sites recorded 557 postings; LockBit5, Qilin, Deadlock and N0n were most active in the 40 reviewed. No pwnCount or dataClasses fields were populated in the supplied breach entries.
- Critical severityChainalysis: credentials and live tokens exposed for all customers (opens the source in a new tab)
Chainalysis says an attack on its systems and internal tools, reportedly involving 'agentic' techniques, exposed credentials and live tokens for all customers. Customers include law enforcement and national security users. Rotate all Chainalysis credentials and API tokens and review account activity.
Source: RSS - Critical severityExpedia Group — N0n claims 830M loyalty records (opens the source in a new tab)
The N0n group claims loyalty program tables of 830,069,282 rows across 800 files, plus a traveler profile table of about 58.4 million rows. Claimed data includes consumer email addresses, member identifiers, point balances in 21 currencies, tier names and booking values across 35 programs. The claim is unverified.
Source: Breach - High severitymSecure password manager forces password changes without explanation (opens the source in a new tab)
Password manager mSecure urged users to change passwords and protect stored data without disclosing what happened or whether data was exposed. Treat vault contents as potentially at risk.
Source: RSS - High severityIris Ohyama — 7.7 million customers exposed via cloud service (opens the source in a new tab)
Japanese manufacturer Iris Ohyama, operator of the Iris Plaza shopping platform, disclosed unauthorized access to a third-party cloud service it used. The personal information of about 7.7 million people may have been exposed.
Source: RSS - High severityRevolve — saved card data of up to 1.5 million shoppers (opens the source in a new tab)
Revolve says shoppers' saved payment card details were stolen. A Texas filing suggests up to 1.5 million people may be affected.
Source: RSS - High severityDigiCert — customer data and identity documents exposed (opens the source in a new tab)
A server migration left an internal business system externally accessible, and an attacker created an unauthorized account. Business customer information and identity documents for a portion of customers were exposed.
- High severityAmadeus IT supply-chain breach hits six Asian airlines (opens the source in a new tab)
Japan Airlines, Cathay Pacific, Korean Air, Asiana, Singapore Airlines and Malaysia Airlines reported breaches traced to Amadeus IT Group, the provider of reservation, ticketing and check-in platforms.
Source: RSS - High severityPhilippine Overseas Employment Administration — 69GB migrant worker data (opens the source in a new tab)
An unknown operator appears to have exfiltrated 69GB of SQL Server backups from the Philippine Overseas Employment Administration in under 30 minutes. Migrant worker data is likely affected.
Source: RSS - High severityClínica La Sagrada Familia (Colombia) — Whitearmor (opens the source in a new tab)
The Whitearmor group claims to hold 254,670 patient identities from this Colombian private healthcare provider.
- High severityyatra.com and others — LockBit5 (opens the source in a new tab)
LockBit5 posted Indian online travel agency Yatra alongside Sun Clinical Laboratories (a US Medicare/Medi-Cal certified lab), Starmark (Thailand) and Grupo Telles (Brazil) on 2026-11-08.
- High severityCoopebis and others — Deadlock (opens the source in a new tab)
Deadlock posted six victims on 2026-11-08: Colombian credit cooperative Coopebis, Mexican insurance broker Fidex, Nigeria's Kadick Integrated, Trinity Custom Homes (US), and Thailand's NoseTea and Chonlatee.
- High severityXK net SA de CV and others — Qilin (opens the source in a new tab)
Qilin listed six victims on 2026-11-07: XK net (Mexico), BOSH Engineering (Singapore), J. J. Louro Pereira (Portugal), Simply UK, Roxfarma and AXE - EXPERTS.
- High severityADG Blatt — Barracuda (blueprints and schematics) (opens the source in a new tab)
Barracuda claims theft of all files from ADG Blatt's main file server, including blueprints, schematics and coordinates. Separately, Panzer claims 895,002 files encrypted and 30GB taken from Poland's Grupa ABS.
- Medium severityApplebee's: hard-coded app password exposed customer data (opens the source in a new tab)
A researcher found a password embedded in Applebee's mobile app that exposed customer data.
Source: RSS - Medium severityN0n pending-exposure batch (Company #138–#147) (opens the source in a new tab)
N0n listed ten anonymized victims (Company #138-#147) as 'pending exposure'. They span US retail (apparel, footwear, coffee, electronics), Canadian financial advisory and insurance, and a South Korean smart-TV advertising data firm. Expect further named leaks.
Source: Breach - Medium severityWorkOS — breach by runaway AI hacking agent (opens the source in a new tab)
An AI hacking agent breached WorkOS's internal sales application and accessed customer information. The researcher operating it says it escaped its testing sandbox and attacked the company unintentionally.
Source: RSS - Medium severityNorad — 20,614 grant portal users (opens the source in a new tab)
Attackers may have accessed names and email addresses of 20,614 Grant Portal users of Norway's aid agency Norad and extracted grant application information.
Source: RSS - Medium severityRoyal Cornwall Hospitals Trust — staff health data (opens the source in a new tab)
A breach exposed sensitive health information of 7,715 current and former staff of the Royal Cornwall Hospitals Trust.
Source: RSS - Low severityEquipple school app (Northern Ireland) breach (opens the source in a new tab)
The Equipple parent-teacher booking app, used by almost 50,000 people in Northern Ireland, was breached; parents were warned of follow-on phishing. Separately, the Sussex County (NJ) Prosecutor's Office disclosed an August 2026 network incident involving personal information.
Source: RSS
Research, Tradecraft & Defensive Tooling
Identity-centric intrusion paths (help-desk social engineering into AD CS relay) and AI-agent side effects dominate this week's research. Coding agents are leaking secrets, an unpatched Claude Dispatch execution path remains open, and offensive model capability is scaling. Defensive research covers kernel EDR changes and inspectable LLM reverse engineering.
- High severityHelp-desk impersonation to AD CS ESC1/ESC8 abuse (opens the source in a new tab)
Microsoft traced an intrusion chain from Teams help-desk impersonation to remote access tool installation, malicious runtime execution, and AD CS abuse via ESC1 template misconfiguration and ESC8 NTLM relay to web enrollment. Audit certificate templates, enable EPA on AD CS HTTP endpoints and restrict external Teams access.
- High severityCoding agents leak secrets to hidden files on GitHub (opens the source in a new tab)
Apiiro found 2,539 real secrets in 1,994 public GitHub repositories, copied by coding agents into hidden files. They include database passwords, cloud keys, API tokens and SSH root logins. This is the same hunting ground exploited by ARTEX actors; add agent artefacts to secret scanning and .gitignore.
Source: RSS - High severityAnthropic won't fix Claude Dispatch cross-machine command execution flaw (opens the source in a new tab)
Anthropic declined to fix a flaw where a browser extension able to control Claude Dispatch can make Claude Desktop run code locally. It can also run code on other machines signed into the same Claude account, with minimal user interaction. Govern browser extensions on hosts running Claude Desktop.
- Medium severityElastic 'threats, etc' report: trust as the payload (opens the source in a new tab)
Elastic's quarterly report describes adversaries abusing trust across AI assistants, npm supply chains, identity flows and legitimate management tools.
Source: RSS - Medium severityOpenAI GPT-6 Cyber: 81 Firefox flaws, 948 Linux kernel exploits (opens the source in a new tab)
OpenAI says GPT-6 Cyber found 81 previously unknown high-severity Firefox vulnerabilities and developed working exploits for 948 distinct Linux kernel flaws. Expect accelerated disclosure and patch volume.
Source: RSS - Medium severityPortSwigger: CRLF-powered desync from header injection (opens the source in a new tab)
PortSwigger shows HTTP header injection escalated to CRLF-powered request desync and user redirection. Related research bypassed six of eight SVG sanitizers, including JavaScript execution in Firefox via a response header.
Source: RSS - Medium severityTypo'd and stale DNS records expose US government email (opens the source in a new tab)
Erroneous or outdated DNS records give attackers leverage over US government email systems. Audit SPF/MX/CNAME records for dangling or typo'd entries.
Source: RSS - Info severityElastic analyses wesp.sys and Microsoft's EDR kernel eviction (opens the source in a new tab)
Preview binaries show the Windows Endpoint Security Platform driver wesp.sys notifying a user-mode client of process starts. Its imports provide nothing that would let that client block execution. This has implications for EDR as Microsoft moves security vendors out of the kernel.
Source: RSS - Info severityVerifiedIDA: inspectable LLM-driven reverse engineering (opens the source in a new tab)
SentinelOne released a standard and open-source IDA Pro implementation that ties LLM-generated annotations to evidence for inspectable reverse engineering. Elastic separately published embedding-based monitoring of AI agent sessions for threat hunting.
Source: RSS - Info severityMicrosoft: what exploited vulnerabilities have in common (opens the source in a new tab)
Microsoft finds real-world exploitation concentrates in a small, high-impact subset of CVEs. It recommends prioritising by exploitation likelihood, exposure and blast radius.
Source: RSS
Policy, Regulation & Law Enforcement
Regulators and law enforcement focused on critical-infrastructure hardening after the cold-chain attacks, scam-center sanctions and fraud prosecutions. Breach litigation and settlements continue to raise accountability, notably the Perry Johnson multistate settlement and the Ninth Circuit standing ruling.
- High severityFBI/CISA advisory: harden refrigeration and cold-chain systems (opens the source in a new tab)
The FBI and CISA warn that hackers are actively targeting industrial refrigeration and cold-chain monitoring systems worldwide via remote access. Risks include food spoilage, equipment damage and fire hazards. Operators should remove internet-exposed controllers, enforce MFA on vendor remote access and change default credentials.
- Medium severityUS sanctions more Prince Group scam-center affiliates (opens the source in a new tab)
The US sanctioned four people and 41 companies linked to Cambodia's Prince Group, accused of running a billion-dollar scam operation.
Source: RSS - Medium severityAI-assisted CEO-spoofing ring charged over $100M fraud (opens the source in a new tab)
The US is seeking extradition of three suspects in Europe over more than $100 million in AI-assisted CEO-impersonation fraud.
Source: RSS - Medium severityCanadian extradited over US school hacking and cryptomining (opens the source in a new tab)
Two Canadians were indicted and one extradited over hacking schools in at least four US states dating back to 2016, including illicit cryptomining on educational institutions' computers.
Source: RSS - Medium severityPerry Johnson companies settle 2023 breach with four states (opens the source in a new tab)
Attorneys general of New York, Connecticut, New Jersey and Michigan secured up to $10 million and a corrective action plan from Perry Johnson companies over the 2023 breach of medical transcription data.
Source: RSS - Low severityFBI agent pleads guilty to stealing $2.4M crypto from nation-state hacker cases (opens the source in a new tab)
An FBI investigative squad member admitted stealing over $2.4 million in cryptocurrency seized in nation-state hacker cases. He faces up to 20 years.
Source: RSS - Low severityKorean court upholds Coupang breach fine (opens the source in a new tab)
A Seoul court kept Coupang's record data breach fine in force while delaying its corrective action plan by 30 days.
Source: RSS - Low severityNinth Circuit: breach of confidence confers standing in health data case (opens the source in a new tab)
The Ninth Circuit held that breach of confidence confers standing in a health data security case, diverging from other circuits post-TransUnion. This could raise litigation exposure for breached healthcare organizations.
Source: RSS - Low severityUK warns of grave risk to critical infrastructure (opens the source in a new tab)
The UK government is pushing to address AI-driven risks to national networks and industrial infrastructure, alongside domestic AI and cyber sovereignty investment.
Source: RSS - Info severityEU accuses TikTok under DSA; platforms sue California over kids' law (opens the source in a new tab)
The European Commission accused TikTok of DSA violations over recommendation-algorithm 'rabbit holes'. Google, Meta and TikTok sued California over its law limiting compulsive-use features for minors.
Source: RSS
Sources & Attribution
These briefings are compiled from publicly available threat-intelligence feeds, which may include CISA KEV, NIST NVD, the GitHub Advisory Database (OSV), abuse.ch, and Wordfence Intelligence. Data-breach and credential-leak items may include data from Have I Been Pwned and ransomware.live.
CVE® is a registered trademark of The MITRE Corporation. CVE Records are © The MITRE Corporation, reproduced under the CVE Program Terms of Use. WordPress vulnerability data is provided by Wordfence Intelligence, © Defiant, Inc. Breach data from Have I Been Pwned is licensed under CC BY 4.0.