Koffei

Daily CyberNews · 12/11/2026

‹ 11/11/202613/11/2026 ›
Today in 30 seconds
  • 🔴 MSPs: review ScreenConnect panel audit logs and extensions; rogue clients pushed SCKit to 288 endpoints.
  • 🔴 Check GitHub repos for wfx branches and short-lived workflow changes; 4,837 hijacked accounts planted a CI secret stealer.
  • 🔴 Disable the N-central Legacy API now: new exploits bypass the hotfix and probing is already seen.
  • 🟠 Patch Apache httpd to 2.4.71 or drop h2/h2c: CVE-2026-119618 allows unauthenticated RCE.
  • 🟠 Patch Ivanti EPM to 2024 SU8: CVE-2026-121653 is a CVSS 9.8 unauthenticated SQLi to RCE.

Vulnerabilities 10

🟠 High impact

Apache HTTP Server mod_http2 race condition allows unauthenticated RCE

CVE-2026-119618CVSS 9.8Apache httpd mod_http2

WhatA race condition in mod_http2 header handling can give unauthenticated attackers code execution on Apache httpd 2.4.55 to 2.4.70 with HTTP/2 enabled.

DoUpgrade Apache httpd to 2.4.71, or remove h2 and h2c from the Protocols directive and keep http/1.1 until patched.

Details and sources
CVE-2026-119618 (CVSS 9.8 per Red Hat) is a race condition in mod_http2's header handling that can give a remote, unauthenticated attacker code execution on servers with HTTP/2 enabled (h2/h2c in the Protocols directive). It affects Apache httpd 2.4.55 through 2.4.70, fixed in 2.4.71 (released Nov 11). The standalone mod_h2 module (2.0.8–2.0.42) has no fixed release and the project is archived. Red Hat ships fixed packages for RHEL 9/10 but rates them "Important" because their per-stream memory allocators make exploitation harder; Debian stable is still unpatched. Exploitation status is disputed: Red Hat says public research shows a reliable exploit but links none, and Wiz reports no public exploit and no in-the-wild use. Not in CISA KEV. Workaround until patched: drop h2/h2c and leave http/1.1. The Hacker News (source)
🟠 High impact

Zoom Workplace for Windows: unauthenticated network privilege escalation

CVE-2026-108186CVE-2026-108187Zoom Workplace Windows

WhatTwo Zoom Workplace for Windows flaws, CVSS 9.6 and 8.1, let an unauthenticated user escalate privileges over the network; no exploitation reported.

DoUpdate Zoom Workplace for Windows to 7.1.8.

Details and sources
CVE-2026-108186 (CVSS 9.6, improper input validation, ZSB-26023) affects versions 7.1.0 to before 7.1.8; CVE-2026-108187 (CVSS 8.1, origin validation error, ZSB-26024) affects versions before 7.1.5. Both let an unauthenticated user escalate privileges over the network. Fixed in 7.1.8 (bulletins of Nov 10); no exploitation reported, EPSS below 1%. The Hacker News (source) · Zoom (source)
🟠 High impact

Ivanti and Citrix November patches: critical EPM SQLi-to-RCE, CSA auth bypass, XenServer pool-secret leak

CVE-2026-121653CVSS 9.8Ivanti EPMCitrix XenServer

WhatIvanti fixed a critical EPM SQL injection to RCE and a CSA auth bypass; Citrix fixed a XenServer pool secret timing leak allowing host takeover.

DoInstall EPM 2024 SU8 and CSA 5.0.8.1, apply Citrix CTX697316 for XenServer, then rotate the pool secret.

Details and sources
Ivanti fixed CVE-2026-121653 (Endpoint Manager, CVSS 9.8, unauthenticated SQL injection leading to RCE; fixed in EPM 2024 SU8), CVE-2026-121683 (Cloud Services Application, CVSS 9.4, missing authentication lets an attacker add or delete trusted EPM Core certificates; fixed in CSA 5.0.8.1) and CVE-2026-105425 (CVSS 7.2, admin-authenticated command injection to RCE in Sentry and EPMM; fixed in Sentry 11.0.0/10.8.3/10.7.4 and EPMM 12.11.0.0/12.10.0.1/12.9.0.3). Ivanti says none is exploited. Citrix's CTX697316 fixes CVE-2026-102705 (High, no CVSS published): an unauthenticated attacker on the management network can infer XenServer 8.4/9's Pool Secret through a timing side channel and take over the host. Patch, then rotate the pool secret. SecurityWeek (source) · Ivanti (source) · Citrix (source)
⚪ Context

CISA KEV: no new additions on Nov 11 (US federal holiday)

CISAKEVno new entries

WhatCISA added no KEV entries on Nov 11 to 12 because of the US federal holiday.

DoFederal agencies must remediate the OpenSSH SSHole, GPUStack, FreeSWITCH and IBM Aspera Faspex KEV entries by today.

Details and sources
CISA's catalog shows no entries dated Nov 11–12 at the time of writing; the newest are the Nov 9–10 batches already covered. Today (Nov 12) is the federal remediation deadline for the OpenSSH "SSHole", GPUStack, FreeSWITCH and IBM Aspera Faspex entries. CISA (source)
🟠 High impact

Node.js November security releases fix 16 flaws, seven high-severity, mostly in HTTP/2 and TLS server code

Added after publishingNode.js16 flawsCVE-2026-123864HTTP/2TLS

WhatNode.js fixed 16 flaws, seven high, including HTTP/2 crashes and use-after-frees and a TLS write bug that can leak server heap memory.

DoUpgrade to Node.js 22.23.5, 24.22.3 or 26.13.1, and move off end-of-life 20.x and 25.x, which stay unpatched.

Details and sources
Node.js 22.23.5, 24.22.3 and 26.13.1 (released Nov 11) fix 16 vulnerabilities. The high-severity ones: CVE-2026-123863 (CVSS 8.2, HTTP/2 teardown assertion lets an unauthenticated client abort the process, 24.x/26.x), CVE-2026-123864 (CVSS 8.1, re-entrancy in zero-length HTTPS TLS writes leaks server heap memory, reportedly including TLS private keys, into response bodies; needs a handler calling res.flushHeaders() then res.end() with data plus a route that echoes request bodies), CVE-2026-123865 (CVSS 7.5, HTTP/2 sessions not cleaned up after a protocol error, leaking memory and file descriptors), CVE-2026-123866 (CVSS 7.5, HTTP/2 stream-teardown use-after-free sends freed memory to the peer), CVE-2026-123874 (CVSS 7.5, HTTP/2 use-after-free in respondWithFile(), 24.x/26.x), CVE-2026-126051 (CVSS 7.5, deferred TLS handshake callback crashes servers using SNI/OCSP/session hooks, 24.x/26.x) and CVE-2026-123872 (CVSS 8.1, TLS 1.2 renegotiation lets a client switch servername and bypass per-host client-certificate rules). The other nine are medium or low: CVE-2026-123877 (6.8, https.Agent reuses TLS sessions across CA policies), CVE-2026-123871 (6.2, proxy credentials in error messages), CVE-2026-123828 (5.9, session.altsvc() heap overflow), CVE-2026-123868 (5.9, HTTP/2 data not counted against maxSessionMemory), CVE-2026-123875 (5.9, cancelled respondWithFile() download crash), CVE-2026-123867 (4.2, SNI client-CA override fallback), CVE-2026-106071 (3.6, wildcard permission rules over-grant), CVE-2026-123862 (2.9, uncatchable zlib assertion) and CVE-2026-123829 (2.5, recursive fs.watch leaks denied file names on macOS/Windows). Scores are the OpenJS CNA CVSS 3.0 ratings in the CVE records; no exploitation or public PoC is reported. End-of-life 20.x and 25.x are affected and unpatched. Five of the flaws were reported by Chris Rohlf with Claude and Anthropic Research. The Hacker News (source) · Node.js (source)
🟠 High impact

PTC patches critical Windchill and FlexPLM deserialization RCE

Added after publishingCVE-2026-126325CVSS 9.2PTC WindchillFlexPLM

WhatA deserialization flaw lets remote unauthenticated attackers run code on Windchill and FlexPLM; no exploitation reported, though an earlier Windchill RCE was exploited.

DoSelf-managed customers should apply the fix from PTC advisory CS483249 and hunt for web shells.

Details and sources
CVE-2026-126325 (CVSS 4.0 9.2 / 3.1 9.0, improper input validation and deserialization of untrusted data) lets a remote, unauthenticated attacker run code through crafted requests on Windchill 11.0 M030 through 13.1.4.0 and FlexPLM 11.0 M030 through 13.1.3.0. PTC has patched its hosted environments; self-managed customers must apply the fix from advisory CS483249 (fixed builds are not listed publicly) and hunt for web shells. No exploitation is reported, but the earlier Windchill RCE CVE-2026-12569 was exploited this summer, including by Cl0p-linked extortion actors. The Hacker News (source) · PTC (source)
🔴 Exploited or active

New N-central exploits bypass N-able's hotfix ("4EverAble2Shell", "Agent2Shell")

Added after publishingN-able N-centralCVE-2026-90436no CVE yet

WhatNew public PoC chains reportedly bypass the N-central 2026.4 HF2 hotfix, and probing chaining an Apache Axis flaw was seen within hours.

DoDisable the N-central Legacy API now and look for probes from 157.20.241[.]16.

Earlier coverage: 11/1110/11
Follow-ups: 13/1118/1119/11
Details and sources
researcher Sean Bales published new PoC chains that reportedly work against N-central 2026.4 HF2 (build 2026.4.1.26), the Nov 10 fix for Able2Shell. 4EverAble2Shell gives unauthenticated root code execution on the server but needs the non-default Legacy API enabled and internet-exposed; Agent2Shell chains 10+ zero-days to run commands as SYSTEM on managed Windows agents when the appliance ID is known. No CVE IDs yet. Presidian saw probing within hours from 157.20.241[.]16, chaining the Apache Axis 1.x unsafe-reflection flaw CVE-2026-90436 (rated critical, Axis is end-of-life, patch released Nov 6) with 4EverAble2Shell, and watchTowr saw probes too; no successful exploitation is confirmed. N-able: disable the Legacy API now; hosted N-central is not affected. The Hacker News (source)
🟡 Worth knowing

AMD Turin (Zen 5) CPUs hit by three Transient Scheduler Attack flaws, Google Cloud warns

Added after publishingAMD Zen 5CVE-2024-36350CVE-2024-36357CVE-2026-0004

WhatAMD Turin CPUs are affected by three Transient Scheduler Attack flaws that can let code on the same core infer cache data.

DoApply OS or hypervisor patches plus microcode; Google Cloud VMs need a full stop and start after the guest fix.

Details and sources
AMD had said Zen 5 was not affected by TSA-SQ CVE-2024-36350 and TSA-L1 CVE-2024-36357 (both CVSS 5.6, local, high complexity) when they were disclosed in July 2025; that statement is now corrected, and a new variant, TSA-Fill (CVE-2026-0004, ID per the Xen Project; record not yet published), can let code on the same core infer L1 data-cache contents. Affected: Google Cloud C4D, N4D, H4D, Z4D and G4 instances and on-premises 5th-gen EPYC servers. Fix: OS or hypervisor patches plus firmware/microcode (in linux-firmware since Sep 23); Google Cloud VMs need a full stop/start after the guest fix, not just a reboot, and performance may drop. Google rates it Medium; no exploitation, not in KEV. The Hacker News (source)
🟠 High impact

Critical pre-auth MongoDB use-after-free crashes replica set primaries

Added after publishingCVE-2026-118383CVSS 9.5MongoDB replica sets

WhatA pre-auth use-after-free lets an unauthenticated client crash a newly elected MongoDB replica set primary; a public PoC exists, no attacks known.

DoUpgrade MongoDB to 9.0.7, 8.3.16, 8.0.37, 7.0.48, 6.0.30, 5.0.35 or 4.4.32.

Details and sources
CVE-2026-118383 (CVSS 4.0 9.5 / 3.1 9.0, CWE-416) lets an unauthenticated client send two ping commands with readConcern.afterOpTime that leave pending waiters; if that lagging member later wins an election, cleanup runs twice and the new primary crashes before it accepts writes. MongoDB says code execution may be possible under some conditions (the researchers did not achieve it). It affects replica set members, including shard and config servers, not standalone mongod; fixed in 9.0.7, 8.3.16, 8.0.37, 7.0.48, 6.0.30, 5.0.35 and 4.4.32, and Atlas is patched automatically. Datadog Security Labs reproduced the crash; a public PoC exists but its reliable mode needs a test-only setting; MongoDB knows of no attacks. The Hacker News (source)
🟠 High impact

Public exploit for unauthenticated RCE in Ruflo's MCP bridge

Added after publishingCVE-2026-59726CVSS 10.0Ruflo MCP bridgepublic exploit

WhatA public exploit targets Ruflo's unauthenticated MCP bridge endpoints, giving a shell in the bridge container and access to provider API keys.

DoUpdate Ruflo to 3.16.3 and keep the MCP bridge off the internet.

Details and sources
Exploit-DB published a remote exploit (EDB-52705, Nov 11) for CVE-2026-59726 (CVSS 10.0): before 3.16.3, Ruflo's default docker-compose deployment leaves the MCP bridge's POST /mcp endpoints unauthenticated, so anyone can call terminal_execute for a shell in the bridge container, read provider API keys and tamper with AgentDB learning-store patterns. Fixed in 3.16.3 (July 2026); CISA rates it automatable with total impact; no exploitation reported, not in KEV. If you run Ruflo, update and keep the bridge off the internet. Exploit-DB (source) · CVE Program (source)

Breaches / Incidents 3

🟠 High impact

Maersk confirms data theft from a legacy system; WhiteArmor demands $5 million

MaerskWhiteArmor$5M demand

WhatMaersk confirmed data theft from one legacy system, while WhiteArmor claims about 100 GB and demands $5 million; the claims are unverified.

Follow-ups: 13/11
Details and sources
Maersk says attackers accessed and extracted data from "one local legacy system" inherited through an acquisition, that its findings "strongly suggest no sensitive data has been disclosed," and that it sees no impact on other systems or customer API/EDI links. The WhiteArmor ransomware gang listed Maersk on its leak site with a 30-day, $5M Bitcoin demand, claiming ~100 GB from the LF Logistics/OCF platform, including driver records with plaintext passwords, 33 super-admin accounts and 27,842 mobile-device logs. The claims are unverified, and Maersk has not confirmed the LF Logistics link. Security Affairs (source) · Cybernews (source)
🔴 Exploited or active

Rogue ScreenConnect clients hit 288 endpoints at 48 organisations in 90 minutes

ConnectWise ScreenConnectSCKitMSP supply chain

WhatAn operator used ScreenConnect's Run command on cloud panels to push a rogue client and SCKit to 288 endpoints at 48 organisations.

DoReview ScreenConnect panel audit logs and installed extensions, especially Labtech Plus and the RMM Plus variants ConnectWise disabled.

Follow-ups: 16/1118/11
Details and sources
Huntress saw an operator use ScreenConnect's Run command feature on cloud-hosted panels from about 17:47 UTC on Nov 11 to push an attacker-controlled client plus the SCKit toolkit to managed endpoints as SYSTEM. ConnectWise says its own infrastructure is not breached and no ScreenConnect vulnerability is involved. As a precaution it disabled five third-party extensions (Labtech Plus, RMM Plus, RMM+, RMM Plus Password Link, RMM+ Password Link) across about 400 partner environments; the initial access vector is still unconfirmed. This is an MSP supply-chain risk: review panel audit logs and installed extensions. Huntress (source) · The Hacker News (source)
🟠 High impact

35,000+ Grok-built apps exposed to anyone with the link

xAI Grok app builderGhostShares35,578 apps

WhatResearchers found 35,578 Grok-built apps reachable without sign-in, about 10% holding business or personal data, because public link sharing was preselected.

Details and sources
Claroc's "GhostShares" research found 37,571 apps built with xAI's Grok app builder, 35,578 of them reachable without sign-in, about 10% holding business or personal data. Examples include a hospital staff list of 3,500+ employees, an ICU patient briefing, plaintext passwords in a contractor timesheet app, and 24 apps built by X/SpaceXSI staff, one a confidential ad proposal to a US federal agency. The cause: "Anyone with the link" was preselected on publish and app hostnames were discoverable through public DNS records. The Hacker News (source)

Malware / Ransomware 5

🔴 Exploited or active

SCKit: worm-capable toolkit built to spread through technicians' ScreenConnect sessions

SCKitScreenConnectworm-capable

WhatThe SCKit toolkit used in the ScreenConnect attacks adds Defender exclusions, hijacks the legitimate client slot and is built to spread via technicians' sessions.

DoHunt for a TrapWatch scheduled task, new Defender exclusions and AppLocker rules on ScreenConnect-managed endpoints.

Follow-ups: 16/1118/11
Details and sources
the 16-file SCKit kit dropped in the ScreenConnect attacks encrypts payloads per host, adds Defender exclusions and AppLocker rules, creates a "TrapWatch" scheduled task that stops the legitimate client, and uses a "RelayTrap" module to take over that client's slot. In code, it is built to steal a connecting technician's .aspxauth session cookie and push itself through the panel's Shared Toolbox to other endpoints, but Huntress has not seen this lateral step succeed. Parts of the code look LLM-generated. Huntress (source) · GBHackers (source)
🔴 Exploited or active

4,837 hijacked GitHub accounts used to plant a CI secret stealer in 1,806+ repos

GitHub ActionsCI secretsGhostAction overlap

WhatAn attacker used 4,837 hijacked GitHub accounts to run a CI secret stealer in at least 1,806 repos, harvesting Actions secrets and cloud keys.

DoCheck repos for wfx-prefixed branches, brief workflow or npm postinstall changes, and use of digitalelvis/neuronai-studio 3.14.3.

Earlier coverage: 29/10
Details and sources
an attacker logged in with stolen developer credentials and pushed short-lived workflow changes, npm postinstall hooks or new "wfx"-prefixed branches. The stealer ran in at least 1,806 repositories, and test PRs reached 61,764 repos (10,431 via collaborator or org access). Changes were usually reverted within a minute and the run logs deleted. It harvests GitHub Actions secrets plus AWS/Azure/GCP and AI-provider keys, and turns OIDC tokens into cloud sessions that survive credential rotation. A Packagist package (digitalelvis/neuronai-studio 3.14.3) and two Go module versions were affected. Accounts overlap with October's GhostAction wave, but Aikido does not attribute it to the same actor. The Hacker News (source) · Endor Labs (source)
🟠 High impact

Citrus ransomware (Sn00py) spreads through malicious Group Policy against Russian firms

Sn00pyToy GhoulsCVE-2026-54121Russia

WhatA new crew spreads Citrus ransomware through malicious Group Policy against Russian firms after entering with stolen contractor credentials over VPN or RDP.

Details and sources
Kaspersky documents a new Russian-speaking crew that overlaps with Toy Ghouls and gets in with stolen contractor credentials over VPN or exposed RDP. It attempted CVE-2026-54121 ("Certighost", AD CS improper authorization, CVSS 8.8, patched July 2026, public PoC) and CVE-2020-1337 (Print Spooler LPE, CVSS 7.8); Kaspersky does not confirm either succeeded. The Go-based Sn00py payload (RSA-2048 + ChaCha20, .citrus extension) is pushed through SYSVOL GPOs that also disable Defender and shadow copies, while domain controllers are skipped so the policy keeps spreading. C2 runs over Xray VLESS/REALITY tunnels disguised as Yandex traffic, with AnyDesk as backup, and Telegram Desktop sessions are stolen before encryption. Securelist (source)
🔴 Exploited or active

Qlik Sense servers hit with reverse shells; ActiveMQ and Veeam exploits staged

Qlik SenseCVE-2023-46604CVE-2024-40711CACTUS

WhatAttackers hit EU Qlik Sense servers with reverse shells and a Go backdoor, staging KEV-listed ActiveMQ and Veeam exploits.

DoHunt for connections to 45.115.26[.]23 and a WinUpdCheck scheduled task on Qlik Sense servers.

Details and sources
Cynet saw EU Qlik Sense Enterprise customers in October 2026 hit with a Nishang PowerShell reverse shell, then a Go backdoor hunting for ActiveMQ, with OpenSSH and a "WinUpdCheck" task for persistence (infrastructure: 45.115.26[.]23). Exploits for CVE-2023-46604 (Apache ActiveMQ, CVSS 10.0, KEV, ransomware use) and CVE-2024-40711 (Veeam Backup & Replication, CVSS 9.8, KEV, used by Akira/Fog/Frag) sat on the attacker's server. The activity partly overlaps with the 2023 CACTUS campaign, which used the Qlik flaws CVE-2023-41265, CVE-2023-41266 and CVE-2023-48365 (all in KEV). Which Qlik CVE was used in 2026 is not confirmed. The Hacker News (source)
🟡 Worth knowing

Remus infostealer v1.3.9 advertises Gmail access that survives a password change

RemusinfostealerGmail claim

WhatThe Remus developer claims Gmail access survives a password reset, but Hudson Rock found no proof or Gmail zero-day.

DoAfter an infostealer infection, revoke sessions rather than only changing the password.

Details and sources
the Remus developer claims stolen sessions keep Gmail access even after the victim resets their password. Hudson Rock found no proof that access persists and no Gmail zero-day, so treat it as an unverified marketing claim; revoking sessions, not just changing the password, is still the right response to an infostealer infection. Malwarebytes (source)

Threat Actor Activity 6

🟠 High impact

AntiFox AI pentesting agents used to pivot from IIS servers into AWS

AntiFoxARTEXAWSIIS

WhatAntiFox AI pentest agents pivoted from an IIS web shell into AWS, cycling stolen IAM credentials to harvest secrets and tamper with S3 objects.

Earlier coverage: 11/1106/11
Details and sources
Arctic Wolf tied Alibaba Cloud hosts running AntiFox, an open-source agentic pentest tool with a Chinese-language interface forked from ARTEX 0.1.1, to real intrusions. The agents went from an ASPX web shell on an IIS server (no CVE named) to config files and connection strings, then cycled stolen IAM users and EC2 roles through hundreds of AWS API calls in about 30 minutes, harvesting Secrets Manager secrets and reading, copying and deleting S3 objects; at least one production secret read succeeded. The actor is not attributed, and AWS itself was not compromised. BleepingComputer (source) · The Hacker News (source)
🟠 High impact

Falcon C2 (XG): browser-based espionage against Middle East and Southeast Asian governments

Falcon C2PeckBirdyAntinogovernments

WhatUnit 42 documents Falcon C2, a browser-based espionage framework injected into Roundcube and government web apps to steal sessions and credentials in 2026.

Details and sources
Unit 42 documents Falcon C2, a Node.js framework that runs JavaScript in victims' browsers. It was loaded through script tags injected into Roundcube webmail (39 instances at one Middle Eastern foreign ministry) and government web apps, plus a fake Flash update serving the Rust Antino implant (OneDrive C2) and a malicious "PDF Viewer" browser extension. It steals cookies and credentials, proxies hijacked sessions, hijacks the clipboard, scrapes Gmail, Discord, Zalo and X, and swaps legitimate downloads for payloads. Targets were defence, interior, foreign affairs, police and justice bodies, March–May 2026. Unit 42 assesses with high confidence that it is a variant of PeckBirdy but names no actor; it cites Talos's China-nexus UAT-11587, which also uses Antino. Unit 42 (source)
🟡 Worth knowing

Romania's SRI formally attributes government intrusion to Turla

TurlaCerealRomania

WhatRomania's SRI formally attributes a December 2025 intrusion at a government organisation to Turla, saying no sensitive data left.

Earlier coverage: 11/1110/11
Details and sources
Romania's intelligence service says Turla (FSB) breached an unnamed Romanian government organisation in December 2025 through "systemic vulnerabilities in Microsoft operating systems" (no CVEs given), aiming at email exchanged with NATO and the EU, and that no sensitive data left. This is official confirmation of the Romanian Cereal case ESET reported on Nov 10. Arab News / Reuters (source)
🟡 Worth knowing

Krebs names the alleged administrator of Qilin ransomware

Qilinalleged administratorOSINT

WhatKrebsOnSecurity names a St. Petersburg man as the alleged Qilin administrator based on open-source evidence; no charges are reported.

Details and sources
KrebsOnSecurity identifies 36-year-old Aleksey Mikhailovich Yakovtsev of St. Petersburg as the person behind the Qilin admin handles "Haise," "Rjun" and "Alexbigboss44." The evidence includes RAMP forum DMs (Intel 471), a mail.ru address tied to Gosuslugi records (Constella), a Telegram ID (Flashpoint) and family contact pivots (IntelHawk). This is open-source investigation, not an official action: no charges have been reported, and Yakovtsev could not be reached. Krebs on Security (source)
🔴 Exploited or active

Operation CloudFire hijacks live Google sessions with calendar invites and a reverse proxy

Added after publishingOperation CloudFireNyasherGoogle Workspace

WhatAn ongoing phishing campaign uses Calendar and Drive invites from known contacts and a reverse proxy to capture MFA and hijack Workspace sessions.

Details and sources
Material Security describes a phishing campaign, active since mid-August, that sends Google Calendar (and now Drive share) invites from compromised accounts the target already knows, routes clicks through trusted Google and Intuit redirectors and a Cloudflare Turnstile gate, then uses the Nyasher reverse-proxy kit to capture passwords and SMS, authenticator and Tap-Yes MFA in real time and take over the Workspace session. Across participating organisations in one month: 68,676 messages blocked, 897 domains taken down and 8,570 compromised senders on 6,354 domains. The operator is unnamed; an unfinished Microsoft 365 branch sits in the code. The Hacker News (source)
🟡 Worth knowing

Scammers abuse the browser "Allow notifications" prompt (Pushpaganda, fake quiz sites)

Added after publishingPushpagandabrowser notificationsscareware

WhatScam sites trick visitors into allowing browser notifications, then push scareware and fraud such as fake arrest warrants long after the visit.

DoRevoke the notification permission for such sites in the browser's site settings.

Details and sources
Malwarebytes warns that sites tricking visitors into allowing notifications keep pushing scareware and fraud long after the visit. HUMAN Security's Pushpaganda operation used AI-generated articles surfaced in Google Discover to get opt-ins, then pushed fake arrest warrants and "Mom" missed-call alerts; quiz sites put "Click 'Allow' to continue" next to the browser prompt. The notification itself installs nothing; revoke the permission in the browser's site settings. Malwarebytes (source)

Law Enforcement 2

🟡 Worth knowing

Delhi Police arrest two alleged leaders of a Jamtara fake-bank-website fraud ring

Delhi PoliceJamtarafake bank sites

WhatDelhi Police arrested two alleged leaders of a ring running look-alike bank websites and fake customer-care numbers.

Details and sources
Delhi Police's cybercrime unit arrested Mohd. Mehruddin Ansari and Mohd. Arsad Ansari, accused of running look-alike bank websites and fake customer-care numbers. One victim lost ₹27.10 lakh, moved through eight mule accounts; eight other members had already been arrested. ANI (source)
🟡 Worth knowing

Former FBI supervisor pleads guilty to stealing $2.47M in crypto from case wallets

Added after publishingFBIinsidercrypto theft

WhatA former FBI supervisor pleaded guilty to stealing $2.47M in crypto from wallets tied to 35 FBI investigations.

Details and sources
Patrick Steven Yaroch, 37, pleaded guilty to wire fraud and money laundering on Nov 5 (reported Nov 12) for about 100 transfers from wallets tied to 35 FBI investigations, some involving nation-state actors: he memorised seed phrases he saw in FBI systems, moved the funds to his own wallet and swapped them to hide their origin. He confessed to a former DOJ colleague and self-reported, and was fired; no sentencing date is reported. Malwarebytes (source) · DOJ (source)
‹ 11/11/202613/11/2026 ›