Daily Security Intelligence Briefing — November 20, 2026

Overall Daily Threat Level: CRITICAL.
Published Friday, November 20, 2026 · 8:01 AM EST
A wave of foundational software vulnerabilities—including a China-linked VMware vCenter zero-day (The Hacker News, SecurityWeek), OpenJDK's 'Java4Shell' (Wiz), NGINX's 'NGINX2Root' (Help Net Security, Cloudflare), and PHP's 'phpAfterShell' (BleepingComputer)—demands immediate global patching. Meanwhile, third-party compromises impacted financial giants JPMorgan, Visa, and Mastercard (The Register), and the European Commission confirmed an AI-driven cyberattack disrupted its HR systems (The Record).

Tier 1 — Priority Targets (Must-Read)

1. China-Linked Hackers Exploit Critical VMware vCenter Zero-Day to Gain Admin Access and Steal Data

Source & Article Link: The Hacker News

Technical Severity / Domain: Virtualization / CVE-2026-123890 (CVSS 9.8, 0-Day)

Why You Must Read This: A China-nexus espionage group tracked as Armory Castle (formerly UNC5221) has been exploiting a critical authentication bypass in VMware vCenter Server's Directory Service as a zero-day since early September 2026. The vulnerability allows unauthenticated attackers with network access to gain administrative control over central virtualization management consoles, exposing entire ESXi and virtual machine fleets. Broadcom has issued emergency patches across vCenter 8.0, 9.0, 9.1, and VMware Cloud Foundation, noting that no workarounds exist.

Recommended Immediate Action: Immediately apply Broadcom's VMSA-2026-0009 patches (8.0 U3m, 9.0.2.0200, 9.1.1.0200), strictly isolate vCenter management interfaces from untrusted networks, and hunt for unauthorized administrative sessions dating back to September 2026.

2. CVE-2026-114938 OpenJDK RCE (Java4Shell): Everything You Need to Know

Source & Article Link: Wiz

Technical Severity / Domain: Enterprise Runtime / CVE-2026-114938 (CVSS 9.2, Active Attempts)

Why You Must Read This: Dubbed 'Java4Shell', this AI-discovered critical flaw in OpenJDK's AVX-512 Base64 decoding routine triggers heap memory corruption and pre-authentication remote code execution. The bug affects OpenJDK 17.0.6 and later running on modern x86-64 hardware, putting huge swaths of enterprise applications—including Jenkins, Confluence, and Apache Tomcat—at risk of unauthenticated takeover. Active exploitation attempts have already begun as cloud providers and software vendors scramble to map vulnerable CPU instruction sets and issue patches.

Recommended Immediate Action: Upgrade OpenJDK runtimes immediately across all production environments, or apply the emergency JVM flag workaround to disable AVX-512 Base64 intrinsics where patching is delayed.

3. NGINX2Root: F5 patches five critical NGINX vulnerabilities

Source & Article Link: Help Net Security

Technical Severity / Domain: Web Infrastructure / CVE-2026-123007 & CVE-2026-123633 (CVSS 9.2)

Why You Must Read This: F5 and nginx.org disclosed ten vulnerabilities in NGINX, including five critical memory corruption flaws dubbed 'NGINX2Root' by researchers. The most severe flaws—an HTTP/2 flow-control integer underflow present since NGINX 1.9.5 (CVE-2026-123007) and an HTTP/2 proxy use-after-free (CVE-2026-123633)—allow unauthenticated remote code execution even on servers only serving static files. Researchers demonstrated chaining these bugs with a Linux kernel SCTP flaw to achieve root privilege escalation and escape Docker containers.

Recommended Immediate Action: Upgrade NGINX Open Source to 1.30.7 or 1.31.9 (and apply corresponding NGINX Plus, cPanel, and Synology updates), or temporarily disable HTTP/2 ('http2 off;') on exposed edge servers.

4. PHP security updates fix 18 flaws, including critical RCE bugs

Source & Article Link: BleepingComputer

Technical Severity / Domain: Web Runtime / CVE-2026-126714 & CVE-2026-129289 (CVSS 9.2)

Why You Must Read This: The PHP Foundation released emergency updates (8.6.0, 8.5.13, 8.4.28, 8.3.37, 8.2.36) addressing 18 vulnerabilities, seven of which are critical remote code execution flaws. Highlights include 'phpAfterShell' (CVE-2026-126714), which abuses interrupted file-based session writes to inject arbitrary objects, CVE-2026-115560 in session deserialization, and a PHP-FPM request smuggling flaw (CVE-2026-129289) allowing unauthenticated RCE when backend connections are reused. Because these impact core PHP session handling and FPM worker pools across all supported branches, millions of web applications face potential compromise.

Recommended Immediate Action: Upgrade PHP runtimes to 8.5.13, 8.4.28, 8.3.37, 8.2.36, or 8.6.0 immediately; if delayed, disable persistent PHP-FPM backend connections and set 'zend.exception_ignore_args=1' in worker configurations.

Tier 2 — Essential Intelligence (Summaries & Analysis)

1. Chase, Visa, Mastercard confirm third-party cyberattacks that may have exposed customers' info

Domain: Third-Party Risk / Financial Services

Subject / Topic Summary: JPMorgan Chase, Visa, and Mastercard have confirmed they are investigating data breaches at third-party service providers after the N0n extortion group leaked alleged customer data on the dark web. Visa noted its exposure is tied to a service provider in the Asia Pacific region, while all three financial giants emphasized their core internal networks remain uncompromised.

Key Technical & Strategic Takeaways: Extortion group N0n claimed responsibility and published customer records allegedly tied to the three financial titans.; The incidents highlight how regional third-party processors remain a primary single point of failure for global financial data.

Essential Direct Quote: “They are associated with an incident involving a third-party service provider in Asia Pacific.”

Source: The Register

2. European Commission confirms cyberattack, says hackers used AI tool

Domain: Public Sector / AI Threats

Subject / Topic Summary: The European Commission confirmed that threat actors breached its HR systems on October 22 using ARTEX, an open-source AI-assisted penetration testing tool recently flagged by the FBI. The intrusion forced the Commission to take several internal applications offline for over a day to contain the threat, though officials state there is currently no evidence of data exfiltration.

Key Technical & Strategic Takeaways: Attackers leveraged the publicly available ARTEX AI hacking framework to automate intrusion steps against EU infrastructure.; The incident validates recent FBI warnings that both state-sponsored and criminal groups are actively operationalizing open-source AI offensive tools.

Essential Direct Quote: “Barely a week after the US Federal Bureau of Investigation (FBI) warned companies of the cyber attacker interest in an open source penetration testing tool called ARTEX, the European Commission has confirmed that attackers armed with that tool broke into its HR systems.”

Source: The Record

3. Exploited Puppet flaw puts the server that configures your endpoint fleet at risk (CVE-2026-115731)

Domain: Endpoint Management / Active Exploitation

Subject / Topic Summary: Perforce has patched two critical vulnerabilities in Puppet Enterprise, including a deserialization flaw (CVE-2026-115731) currently under active exploitation in the wild. The vulnerability allows unauthenticated attackers to execute arbitrary code and impersonate managed devices, potentially handing adversaries full control over an organization's automated endpoint fleet.

Key Technical & Strategic Takeaways: CISA has added CVE-2026-115731 to its Known Exploited Vulnerabilities (KEV) catalog.; Vox Pupuli also disclosed a related factset overwrite issue in openvoxdb-termini.

Essential Direct Quote: “These flaws allow unauthenticated attackers to execute arbitrary code and impersonate managed devices, potentially compromising an entire organization's endpoint infrastructure.”

Source: Help Net Security

4. Ruby Patches Ruby2Shell Vulnerability That Could Lead to RCE

Domain: Application Security / Vulnerability

Subject / Topic Summary: Ruby maintainers patched 'Ruby2Shell' (CVE-2026-132825), a critical heap-based buffer overflow triggered when upcasing strings in single-byte encodings (such as the German 'ß' character expanding from one byte to two). The flaw has existed since Ruby 2.4.0 and enables remote code execution on stock Rails deployments, prompting urgent updates for self-hosted platforms like GitLab (patched in 19.6.1) and Mastodon (patched in 4.7.6).

Key Technical & Strategic Takeaways: UTF-8 strings are unaffected, but single-byte encodings trigger a 1-byte overflow that corrupts heap memory.; A public proof-of-concept exploit is scheduled for release on November 25, making immediate upgrades or WAF virtual patching critical.

Essential Direct Quote: “Tracked as CVE-2026-132825, the bug is a heap-based buffer overflow that could lead to remote code execution (RCE).”

Source: SecurityWeek

5. F5 releases emergency patches for critical BIG-IP RCE flaws

Domain: Network Security / Vulnerability

Subject / Topic Summary: F5 Networks issued emergency patches for five critical vulnerabilities in BIG-IP appliances that allow remote attackers to trigger denial-of-service conditions or achieve remote code execution. Given BIG-IP's frequent placement at the network perimeter, security teams should prioritize patching before exploit code surfaces.

Key Technical & Strategic Takeaways: Five critical flaws were addressed across BIG-IP modules, carrying risks of full system compromise or service disruption.; Edge appliances remain prime targets for automated scanning and rapid weaponization following patch releases.

Essential Direct Quote: “US cybersecurity company F5 warned customers to patch five critical BIG-IP vulnerabilities that attackers can exploit to trigger denial-of-service (DoS) or gain remote code execution.”

Source: BleepingComputer

6. Hackers Exploit Critical Cisco Expressway Flaw, Breach Four Devices and Reach Internal Network

Domain: Network Security / Active Exploitation

Subject / Topic Summary: Threat actors exploited a recently disclosed critical vulnerability in Cisco Expressway (CVE-2026-76466, CVSS 9.8) shortly after a proof-of-concept exploit became public. Attackers compromised four clustered Expressway gateways at a Russian organization and deployed lateral movement tooling to pivot into the internal corporate network.

Key Technical & Strategic Takeaways: Successful exploitation leaves a distinct 'nodeup, wombat@127.0.0.1' entry in the device's Erlang system log (erlang.log).; Unauthenticated remote attackers with network access can fully compromise affected Cisco Expressway clusters.

Essential Direct Quote: “One of the key indicators of successful exploitation of the vulnerability is the appearance of the entry nodeup,'wombat@127.0.0.1' in the Erlang system log (erlang.log) on Cisco Expressway devices.”

Source: The Hacker News

7. ACritical OpenSSL Vulnerability (CVE-2026-121481) Disclosed in EVP_EncodeBlock()

Domain: Cryptography / Vulnerability

Subject / Topic Summary: OpenSSL patched a CVSS 9.8 heap buffer overflow (CVE-2026-121481) in the EVP_EncodeBlock() Base64 encoding function affecting versions 4.0.0–4.0.3 and 4.1.0. When processing massive inputs (~1.5GB or greater), the output size calculation overflows a 32-bit integer, causing out-of-bounds memory writes; it is reachable via CLI utilities like 'openssl s_client -proxy_pass' with oversized passwords.

Key Technical & Strategic Takeaways: Fixed in OpenSSL 4.0.4 and 4.1.1 by returning -1 when encoded sizes overflow an integer.; While rated 9.8 due to potential network reachability, practical exploitation requires an application to pass ~1.5GB of data into a single Base64 encode call.

Essential Direct Quote: “The flaw involves an integer overflow when calculating the encoded output size for input lengths approaching INT_MAX.”

Source: #cve / @thecybersecguru

8. A single pull request could have handed attackers the keys to Microsoft’s software supply chain

Domain: Supply Chain Security / Windows

Subject / Topic Summary: Cycode researchers discovered a now-patched critical zero-click supply chain vulnerability in the open-source Windows Package Manager (WinGet) repository pipeline. By submitting a single malicious pull request, an attacker could have hijacked an automated validation script running with privileged credentials, potentially compromising software distribution to billions of Windows devices.

Key Technical & Strategic Takeaways: Automated CI/CD workflows processing untrusted pull requests with high-privilege tokens remain a top supply chain vector.; Microsoft patched the pipeline flaw before any malicious packages or token thefts occurred.

Essential Direct Quote: “The most dangerous actor in a software supply chain may now be an automated script with powerful credentials and no idea where it’s sending them.”

Source: CSO Online

9. Russian spies turn one cloud breach into a stepping stone for the next

Domain: Threat Intelligence / Cloud Identity

Subject / Topic Summary: Google Threat Intelligence Group (GTIG) reports that Russian FSB-linked espionage group COLD RELIC (formerly COLDRIVER / Star Blizzard) breached a US organization's Google Workspace using stolen Super Admin credentials and MFA push fatigue. Even after the victim reset passwords and sessions, the attackers maintained persistent access for months because they had registered rogue passkeys, backup codes, and MFA devices during initial access.

Key Technical & Strategic Takeaways: Password and session resets are insufficient for cloud account takeover recovery if attackers register persistent passkeys or OAuth grants.; COLD RELIC uses compromised organizations to impersonate trusted contacts and phish downstream government and defense targets.

Essential Direct Quote: “The victim reset passwords and sessions. The passkeys survived, so access lasted months. Account takeover cleanup = revoke passkeys, backup codes, MFA devices and OAuth grants.”

Source: Help Net Security

10. Drupal Patches Critical Installer Flaw That Could Give Unauthenticated Attackers Control of Sites

Domain: Web Applications / Vulnerability

Subject / Topic Summary: The Drupal Security Team patched a critical remote code execution vulnerability (CVE-2026-132165) in Drupal core that allows unauthenticated attackers to access install.php, restart the installation process, and overwrite settings.php with attacker-controlled database credentials. While vulnerable configurations are considered uncommon, the flaw grants full administrative takeover and code execution, and end-of-life branches receive no fix.

Key Technical & Strategic Takeaways: Upgrade to Drupal 11.4.13 or latest supported branch releases immediately.; As an interim mitigation or for EOL branches, block network access to install.php and enforce read-only permissions (chmod 444) on settings.php.

Essential Direct Quote: “An unauthenticated attacker reaches install.php, restarts the install process, and overwrites settings.php with their own database credentials.”

Source: The Hacker News

11. SonicWall Warns Gen 6 Firewalls Won't Get Security Patches, Even for Critical Flaws

Domain: Network Security / Lifecycle Management

Subject / Topic Summary: SonicWall issued an urgent warning advising customers to immediately retire and replace all Generation 6 firewalls running SonicOS 6.x and remove them from the internet edge. The vendor emphasized that these end-of-life devices will no longer receive patches even for critical zero-days, noting that adversaries are using AI and automation to rapidly discover and exploit unsupported perimeter hardware.

Key Technical & Strategic Takeaways: SonicOS 6.x hardware is permanently out of security support regardless of flaw severity.; Internet-facing EOL firewalls face imminent automated exploitation and must be migrated to modern hardware immediately.

Essential Direct Quote: “Attackers now use automation and AI to find and exploit unsupported edge devices faster than ever. We expect internet-facing Gen 6 firewalls to be actively targeted.”

Source: The Hacker News

12. N-able Blames Service Disruption on Cyberattack

Domain: MSP Security / Incident Response

Subject / Topic Summary: IT management software provider N-able confirmed that an ongoing service outage affecting its Mail Assure and SpamExperts Hosted Cloud platforms was caused by a cyberattack. While underlying mail flow and filtering continue to operate, N-able took dashboard, API, and quarantine access offline to contain the breach and warned that credential theft may have occurred.

Key Technical & Strategic Takeaways: MSP and email security cloud consoles remain high-value targets for supply chain and credential harvesting attacks.; Customers using Mail Assure and SpamExperts Hosted Cloud should rotate associated credentials and API keys.

Essential Direct Quote: “The incident affected Mail Assure and SpamExperts Hosted Cloud and could have led to credential theft.”

Source: SecurityWeek

13. ASUS warns of critical hard-coded key flaw in Control Center Express

Domain: Enterprise Management / Vulnerability

Subject / Topic Summary: ASUS released Control Center Express v1.7.29 to address three critical vulnerabilities, the worst being a CVSS 10.0 hard-coded cryptographic key (CVE-2026-114202) that grants unauthenticated remote administrative access. Two additional flaws (CVE-2026-114204 and CVE-2026-114205) allow session token forgery and privilege escalation to admin across the IT management console.

Key Technical & Strategic Takeaways: Unauthenticated attackers with network access to the ASUS management console can take full control of managed endpoints and servers.; Upgrade to ASUS Control Center Express v1.7.29 immediately and restrict console access to isolated management VLANs.

Essential Direct Quote: “Worst is CVE-2026-114202 (CVSS 10.0), a hard-coded crypto key granting unauthenticated remote admin.”

Source: BleepingComputer

14. New Unpatched Windows 0-Day FastUnlock Lets Someone Holding a Locked PC Run Programs on It

Domain: Endpoint Security / Windows 0-Day

Subject / Topic Summary: Security researchers disclosed 'FastUnlock', an unpatched Windows 11 zero-day technique with public proof-of-concept code that abuses the Narrator accessibility feature on the lock screen to launch arbitrary programs without signing in. While the bypass requires physical access and only grants a limited user token rather than administrator privileges, it allows attackers to interact with applications on locked, BitLocker-enabled workstations.

Key Technical & Strategic Takeaways: Requires physical access to a locked Windows 11 machine and yields non-elevated user execution.; Organizations worried about physical tampering should disable lock-screen accessibility shortcuts via Group Policy or enforce hibernation/shutdown when unattended.

Essential Direct Quote: “A newly disclosed unpatched zero-day technique called FastUnlock lets someone with physical access to a locked Windows computer start programs without signing in.”

Source: The Hacker News

15. BabidiGhost: An AI-Augmented Cybercrime Operation

Domain: Threat Intelligence / Offensive AI

Subject / Topic Summary: SOCRadar researchers uncovered 'BabidiGhost', a French-speaking threat actor who spent nine days targeting European banks, insurers, payment providers, and municipalities by commanding a jailbroken AI agent in plain, misspelled French. Meanwhile, Symantec reported a separate AI-authored 41-module PowerShell toolkit used to test stolen credentials and API keys against a major African stock exchange, and Cisco Talos detailed North Korea-aligned UAT-12643 hiding Overlord RAT C2 on the Solana blockchain.

Key Technical & Strategic Takeaways: Low-skill operators are delegating end-to-end technical recon, exploit generation, and data extraction to jailbroken AI agents.; AI-generated offensive toolkits exhibit distinct structural modularity, while nation-state actors continue innovating resilient C2 channels like public blockchains.

Essential Direct Quote: “The operator wrote his orders in plain, misspelled French and left the technical work to the agent.”

Source: SOCRadar

16. Your AI red team may be finding the same prompt injection thousands of times

Domain: AI Security / Research

Subject / Topic Summary: Academic researchers from the University of Waterloo, Zhejiang University, and KTH Royal Institute of Technology found that retraining AI agents on massive datasets of automated prompt injection attacks actually made the models easier to hijack in three out of four benchmarks. Alongside this, a Postman survey of 9,000 developers revealed that 25% of organizations already run AI agents in production, often without proper API key scoping or governance.

Key Technical & Strategic Takeaways: Automated red-teaming tools often generate thousands of superficial variations of the same prompt injection, causing defensive overfitting that weakens overall model resilience.; Enterprise AI agent deployments in production are outpacing basic API credential management and tool-call interception controls.

Essential Direct Quote: “The automated red-teaming tool that found more working attacks produced the weaker defense.”

Source: Help Net Security

17. Critical Enterprise Application Advisories: Oracle EBS, Spring, Tomcat, Okta, and WatchGuard

Domain: Vulnerability Management / Enterprise Apps

Subject / Topic Summary: A wave of high-severity vendor advisories dropped alongside the major runtime patches: Oracle E-Business Suite patched pre-auth RCE chain CVE-2026-114924 and CVE-2026-114925 (CVSS 9.8); Spring patched 50 flaws including CVSS 9.8 RCE CVE-2026-59273; Apache Tomcat addressed a critical TLS 1.3 authentication bypass (CVE-2026-133325); and CISA added four actively exploited bugs (Apache Axis CVE-2026-90436, LiteLLM CVE-2026-40217, Synology DSM CVE-2024-10441, and REDCap CVE-2026-90817) to its KEV catalog. Additionally, Okta fixed 8 flaws, WatchGuard Dimension 2.3.2 fixed 15 flaws including critical RCE CVE-2026-86130, Shopware patched a CVSS 9.8 SQLi, and Frappe warned of an actively exploited authorization bug.

Key Technical & Strategic Takeaways: Prioritize CISA KEV additions (Apache Axis, LiteLLM, Synology DSM, REDCap) and actively exploited Frappe ERPNext instances.; Schedule urgent maintenance windows for Oracle EBS (12.2.3–12.2.15), Spring Framework, and TLS 1.3-enabled Apache Tomcat clusters.

Essential Direct Quote: “Oracle E-Business Suite flaws CVE-2026-114924 and CVE-2026-114925 (CVSS 9.8) chain into no-login code execution. Affects EBS 12.2.3 to 12.2.15, internal-only installs too.”

Source: #cve / @suriq

Tier 3 — Industry Noise & Awareness Only

• Transport for London Breach — Breach exposes personal data and security answers of 265,000 Oyster photocard holders. (Help Net Security)

• SBC Medical Group Breach — Hackers steal sensitive personal data of 1.6 million customers across two incidents. (BleepingComputer)

• Nvidia Ransomware Claim — Nvidia investigates possible third-party breach after Gentlemen ransomware gang claims hack. (The Register)

• Second Norwegian Energy Breach — Second Norwegian energy company breached in two days amid heightened sabotage warnings. (Dark Reading)

• PavinLoader & ROMULUS ClickFix — PavinLoader uses trojanized phone library and custom VM; ROMULUS delivers CNCMachineRMS RAT. (The Hacker News)

• CyberPower PowerPanel Abused — Attackers abused CyberPower UPS management software to execute malicious scripts on a domain controller. (Infosecurity Magazine)

• Linux Kernel NoisyTCP (CVE-2026-99288) — 11-year-old Linux kernel TCP vulnerability triggers remote kernel panic and denial of service. (#cve / @cybersecurefox)

• Apple macOS & iMessage Updates — Apple tightens macOS Full Disk Access against AI agents and patches iMessage group chat flaws. (#infosec / @knoppix95)

• UK Secret Apple Encryption Notice — UK tribunal confirms British government served Apple with a secret notice targeting iCloud encryption. (The Record)

• US National Cyber Director & CISA Nominee — Sean Cairncross exits as National Cyber Director; CISA nominee Michael LoSapio faces contentious Senate hearing. (CyberScoop)

• DOJ & DHS Threat Disruption Center — DOJ and DHS partner on new unit to triage cybercrime complaints and disrupt scam networks. (The Record)

• Ukraine Arrests Ransomware Affiliates — Ukrainian police arrest four members of a ransomware network that extorted $40M from US firms. (The Record)

• Russian Spy Indicted in Silicon Valley — Russian national indicted for physically stalking and shoulder-surfing Palo Alto defense-tech employees. (Ars Technica)

• AI Math Advances vs. PKI — Cryptographer Matthew Green sparks debate warning AI mathematical advances could threaten public-key cryptography. (CSO Online)

• Android Global Developer Verification — Google will enforce developer identity verification across all Android install sources globally starting January 2027. (Help Net Security)

• Windows 11 Passkey Bug & Developer Telemetry — Microsoft fixes false passkey failure message on Android; researchers note Windows 11 logs coding agent usage. (The Register)

• Microsoft Defender MPenginedb Forensics — Technical walkthrough details how to extract hidden forensic telemetry from Defender's MPenginedb database. (#infosec / @r1cksec)

• Webworm Custom SoftEther Bridge & Timeroasting — Rapid7 details Webworm's malware-free SoftEther espionage bridge; Huntress shares Timeroasting detection methods. (#infosec / @AAKL)

• APFS Unicode Collisions & Plugin CVEs — Unicode normalization flaws hit Deno, setuptools, and libarchive; critical flaws patched in Gladys Assistant and JTL-Shop. (#cve / @cybersecurefox)

• Careers360 & Godot & Fils Breaches — 3.6M accounts leaked from Careers360; French gold retailer Godot & Fils customers robbed after breach. (#infosec / @infonewscyber)

• PostgreSQL 19, Restic 0.20.0 & Dependency-Track 5.2.2 — PostgreSQL 19 ships after two-month delay from AI bug reports; Restic and OWASP Dependency-Track release security fixes. (Simon Willison's Weblog)

• AWS mTLS for NLB & WAF Analytics Dashboard — AWS adds mutual TLS client authentication to Network Load Balancers and launches a new WAF security analytics dashboard. (AWS Security Blog)

• Google Chrome 'Mirror Dimension' & VRP Policy Update — Google builds synthetic web environment to test browser AI agents safely and updates VRP disclosure rules for AI bug hunting. (Google Security Blog)

• Cloudflare 2026 US Midterm Election Traffic Report — Cloudflare reports a 363% election day traffic surge and notes attackers shifted from DDoS to internal system intrusion. (The Cloudflare Blog)

• AI Platform Token Fraud & Self-Hosted Qwen Risks — ChatPRD founder details massive token fraud waves; Dark Reading highlights steering failures in self-hosted Qwen AI agents. (Simon Willison's Weblog)

• Honeypot Reports: Booking.com Phishing & Exposed Docker/Redis — Researchers observe stolen SMTP credentials firing 285 Booking.com lures in 25 seconds and exposed Docker APIs compromised in 5 hours. (#infosec / @lelabtechnique)

• Out-of-Band Network Resilience & SOC AI Governance Surveys — Opengear finds few CISOs have immediate out-of-band network recovery access; Elastic finds 96% of SOCs use AI but only 18% enforce governance. (Help Net Security)

• Ubuntu Linux Kernel, OpenJDK, NGINX & Ruby Security Notices — Canonical issues multiple USN advisories patching Linux kernel, OpenJDK/CRaC JDK, NGINX, Ruby, and PyJWT packages. (Ubuntu security notices)

• Splunk Enterprise & SOAR CVE Alerts — RedPacketSecurity flags over a dozen newly published CVE advisories across Splunk Enterprise, Enterprise Security, and SOAR. (#cve / @RedPacketSecurity)

• Navigating AI Vulnerability Hysteria & cURL Rust Rewrites — Dark Reading, Cisco Talos, and cURL's Daniel Stenberg reflect on warning fatigue and the flood of AI-generated vulnerability reports. (Dark Reading)

← View past briefings (History)