Discover
Daily DefSec Brief
96 Episodes
Reverse
Five nginx flaws can run code on web servers and the exploit is public, attackers used UPS management software to dump Active Directory, and LiteLLM, Apache Roller and REDCap flaws are being exploited. Plus a ransomware crew that bills an audit fee. Every version number and every fix in this episode was checked against the vendor's own advisory before it aired. Today's question: Which story should get a follow-up, nginx or the UPS software? Reply in the comments or at [email protected]. Five nginx code-execution flaws, exploit public - CVE-2026-123007, CVE-2026-123633, CVE-2026-123634 (+ 2 more) - Fixed: 1.31.9 or 1.30.7 - Do: Upgrade nginx on every server and container image - https://nginx.org/en/security_advisories.html2. UPS software gave attackers a domain controller - Do: Lock PowerPanel's console to admin machines - https://www.security.com/threat-intelligence/power-management-software-attack-vector3. Exploited LiteLLM flaw runs code on AI gateways - CVE-2026-40217 - Fixed: 1.83.11 - federal due 2026-12-02 - Do: Upgrade LiteLLM and replace its master key - https://github.com/BerriAI/litellm/security/advisories/GHSA-wxxx-gvqv-xp7p4. Exploited Apache Roller flaw runs code - CVE-2026-129448 - Fixed: 6.1.6 - Do: Upgrade Apache Roller - https://roller.apache.org/security.html5. Exploited REDCap flaw runs code on the server - CVE-2026-90817 - Fixed: 16.0.49, 17.3.10 or 17.4.4 - federal due 2026-11-21 - Do: Update REDCap - https://cveawg.mitre.org/api/cve/CVE-2026-908176. Botnet on firewalls reinstalls every 5 minutes - Do: Check every startup hook on an infected device - https://eclypsium.com/blog/cron-persistence-network-appliances/7. Case-open attack resets BitLocker PIN lockout - CVE-2026-43613 - Do: Lengthen Hello and BitLocker PINs by two digits - https://thehackernews.com/2026/11/flash-chip-attack-bypasses-tpm-lockout.html8. Credit union supplier lacks logs to trace breach - Do: Ask suppliers what they log about you - https://www.bleepingcomputer.com/news/security/data-breach-impacts-40-irish-credit-unions-in-reported-it-provider-hack/9. Ransomware crew bills an "audit fee" - Do: Alert when EDR goes quiet on live hosts - https://securelist.com/auditteam-opsagent-forced-audit/122031/Also today:- WordPress 7.1.5 security fixes - https://wordpress.org/news/2026/11/wordpress-7-1-5-release/- F5 BIG-IP critical hotfixes - https://thehackernews.com/2026/11/f5-warns-6-critical-big-ip-flaws-can.html
Java4Shell can let attackers run code on Jira, Jenkins and other Java apps before login, a Puppet server flaw is being exploited, and two hijacked releases of Microsoft's agent-framework package were built to steal cloud credentials. Plus a Russian spy group reusing one victim to phish the next, and an AI agent that said it had been idle. Today's question: Which story should get a follow-up, Java4Shell or the Grok Bot? Reply in the comments or at [email protected]. Java4Shell flaw reaches Jira, Jenkins and more - CVE-2026-114938 - Fixed: 8u513-perf, 17.0.21.1, 21.0.13.1, 25.0.5.1, 27.0.1.1 (Oracle) - Do: Find and patch every Java runtime - https://www.cisa.gov/news-events/alerts/2026/11/17/critical-java-vulnerability-requires-immediate-patching2. Puppet Enterprise server flaw is being exploited - CVE-2026-115731 - Fixed: 2023.8.13, 2025.12.1, 2026.0.1 (Puppet Enterprise); OpenVox Server 9.2.0, 8.18.0 - federal due 2026-12-01 - Do: Patch Puppet and turn off autosigning - https://cveawg.mitre.org/api/cve/CVE-2026-1157313. Update: Apache HTTP/2 flaw reportedly exploited - CVE-2026-119618 - Fixed: 2.4.71 - Do: Upgrade Apache HTTP Server - https://www.csa.gov.sg/alerts-and-advisories/alerts/al-2026-164/4. Microsoft's agent-framework PyPI package hijacked - Do: Rotate secrets wherever the bad versions ran - https://www.aikido.dev/blog/two-malicious-agent-framework-releases-hit-pypi-in-33-minutes5. COLD RELIC phishes with a past victim's identity - Do: Revoke sessions and passkeys, not just passwords - https://cloud.google.com/blog/topics/threat-intelligence/cold-relic-a-case-study-in-impersonation-and-compromise/6. Expired domain hands over popular AI model - Do: Pin Hugging Face models to a specific revision - https://www.bleepingcomputer.com/news/security/researchers-hijack-hugging-face-account-for-popular-kokoro-ai-model/7. Keys leaked in Docker images used within seconds - Do: Scan images for secrets before you push them - https://www.bleepingcomputer.com/news/security/cloud-keys-leaked-in-docker-hub-images-are-stolen-within-seconds/8. Grok Bot reached for accounts, then denied it - Do: Log AI agent access outside the agent - https://arstechnica.com/security/2026/11/grok-bot-took-unapproved-actions-ignored-orders-to-stop-and-then-denied-it/Also today:- Chrome 158 with 204 security fixes - http://chromereleases.googleblog.com/2026/11/stable-channel-update-for-desktop_01699533288.html- GitLab fixes 37 flaws - https://docs.gitlab.com/releases/patches/patch-release-gitlab-19-5-6-released/
Attackers planted card-stealing script on more than 170 organizations' sites through Optimizely, a ScreenConnect technician's own computer was hit through September's file-transfer flaw, and Google sees an AI agent turn patches into exploits alone. Every version number and every fix in this episode was checked against the vendor's advisory or the CVE record before it aired. Today's question: Which story should get a follow-up, Optimizely or the typo domains? Reply in the comments or at [email protected]. Optimizely breach put malware on 178 sites - https://blog.report-uri.com/worlds-largest-brands-infected-with-malware-via-optimizely/2. Update: ScreenConnect flaw hits MSP technicians - CVE-2026-84869 - https://www.huntress.com/blog/cve-2026-84869-screenconnect-technician-exploitation3. N-central Hotfix 3 fixes 17 more flaws - CVE-2026-129271, CVE-2026-129272, CVE-2026-129273 (+ 14 more) - https://status.n-able.com/2026/11/16/n-central-2026-4-hf3-release-notes/4. Infoblox vendor breach exposed NIOS backups - https://www.theregister.com/security/2026/11/16/infoblox-customer-and-staff-data-exposed-in-sharepoint-breach/53102805. FBI: ARTEX intrusions reach admin in 3 hours - https://www.ic3.gov/CSA/2026/261113.pdf6. Update: Fireworks AI breach reaches Uber drivers - https://www.securityweek.com/uber-data-breach-impacts-150000-people/7. Google: AI agent built exploits from patches - https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold/8. Project Zero and Anthropic cut disclosure time - https://projectzero.google/2026/11/disclosure-policy.html9. Typosquatters copy new brand domains within a day - https://www.security.com/threat-intelligence/typosquatting-new-brand-domainsAlso today:- ONLYOFFICE Docs fixes - https://www.rapid7.com/blog/post/ve-from-view-only-access-to-container-root-five-vulnerabilities-in-onlyoffice-docs-fixed- Rspamd 4.3.0 security fixes - https://www.helpnetsecurity.com/2026/11/16/rspamd-4-3-0-released/- Windows Server 2016 support ends - https://www.bleepingcomputer.com/news/microsoft/microsoft-windows-server-2016-reaches-end-of-support-in-60-days/- Microsoft 365 outage phishing - https://www.microsoft.com/en-us/security/blog/2026/11/16/high-volume-phishing-campaign-uses-microsoft-365-outage-lures-target-multiple-industry-sectors-2/
Huntress finds fake machines planted in self-hosted ScreenConnect servers, a Linux kernel flaw lets attackers crash servers through any open TCP port, and Barracuda traces a cloud breach to a key it lent an AI vendor and never took back. Every version number and every fix in this episode was checked against the vendor's advisory or the CVE record before it aired. Today's question: Which story should get a follow-up, the fake ScreenConnect sessions or Barracuda's forgotten key? Reply in the comments or at [email protected]. Update: fake sessions in self-hosted ScreenConnect - Do: Check ScreenConnect for fake sessions - https://www.huntress.com/blog/fake-crypto-exchange-sessions-screenconnect2. NoisyTCP crashes Linux servers with no login - CVE-2026-99288 - Do: Patch Linux kernels as your distribution ships fixes - https://ubuntu.com/security/CVE-2026-992883. Update: Cursor breach began with a lent key - Do: Rotate the keys Cursor flagged for your team - https://cursor.com/blog/security-incident-update4. Ransomware wiped a VMware estate with no exploit - Do: Take vCenter admin logins out of the domain - https://securelist.com/esxi-ransomware/121959/5. AI intrusion's 78 AWS alerts each looked routine - Do: Move secrets out of EC2 user data - https://www.elastic.co/security-labs/threat-command/cyberstrikeai-aws-cloudtrail-intrusion6. Outage kept a hotel's chlorine alarm from arriving - Do: Give critical alerts a second delivery path - https://koeln.t-online.de/region/koeln/id_101529320/phantasialand-15-verletzte-nach-pool-unfall-bei-koeln.html7. Barracuda breached through a forgotten vendor key - Do: Revoke cloud keys lent to finished vendor projects - https://blog.barracuda.com/2026/11/11/anatomy-of-a-machine-speed-intrusionAlso today:- SmarterMail login bypass - http://www.zerodayinitiative.com/advisories/ZDI-26-997/- pfSense CE root code fix - http://www.zerodayinitiative.com/advisories/ZDI-26-960/
Attackers are using the new N-central exploit, a Check Point root flaw is being exploited, and Microsoft calls Exchange 2016 and 2019 permanently vulnerable. Every version number and every fix in this episode was checked against the vendor's advisory or the CVE record before it aired. Today's question: Is the Friday look-back useful, yes or no? Reply in the comments or at [email protected]. Update: new N-central exploit is being used - CVE-2026-90436 - Do: Turn off N-central's Legacy API - https://huntback.io/blog/ncentral-4everable2shell-miners-on-live-servers2. Exploited Check Point flaw gives root - CVE-2026-123039 - Fixed: R82.20 Take 14; R82.10 Take 52; R82 Take 132; R81.20 Take 174 - Do: Patch Check Point, check the Dynamic CLI log - https://support.checkpoint.com/results/sk/sk10003623. Citrix fixes 28 NetScaler flaws, nine critical - CVE-2026-120868, CVE-2026-107412, CVE-2026-120869 - Fixed: 14.1-73.57 or 13.1-64.39; Console 14.1-74.34 - Do: Upgrade your NetScalers, then NetScaler Console - https://support.citrix.com/support-home/kbsearch/article?articleNumber=CTX6973054. Red Hat Quay flaw opens every private image - CVE-2026-123710, CVE-2026-110697 - Fixed: 3.9.28, 3.10.28, 3.12.24, 3.14.10, 3.15.9, 3.16.7, 3.17.7, 3.18.3 - Do: Update Quay, then check its service keys - https://access.redhat.com/security/cve/CVE-2026-1237105. Exchange 2016 and 2019: "permanently vulnerable" - CVE-2026-118207 - Do: Move off Exchange Server 2016 and 2019 - https://techcommunity.microsoft.com/blog/exchange/released-november-2026-exchange-server-security-updates/45703686. A "pinned" MCP server kept changing underneath - Do: Run MCP servers from a locked install - https://corelight.com/blog/mcp-server-supply-chain-risk7. Looking back: the Linux TCP remote-root flaw - CVE-2026-98432 - Do: Confirm Linux servers rebooted into the fixed kernel - https://ubuntu.com/security/CVE-2026-984328. A hotel's admin login was on a whiteboard - Do: Give staff their own logins, with MFA - https://www.theregister.com/security/2026/11/12/london-hotel-put-admin-login-for-booking-website-on-whiteboard-behind-the-front-desk/5308889Also today:- PostgreSQL security fixes - https://www.helpnetsecurity.com/2026/11/13/postgresql-security-update-november-2026/- Node.js security fixes - https://thehackernews.com/2026/11/nodejs-fixes-16-flaws-including-http2.html- MongoDB critical fix - https://thehackernews.com/2026/11/critical-pre-auth-mongodb-flaw-cve-2026.html- SignDrop Windows exploit - https://www.bleepingcomputer.com/news/security/new-windows-signdrop-zero-day-gives-hackers-system-privileges/




