Happy Friday, Security Gang,
This week’s threat landscape made one thing unmistakably clear: the attack surface is no longer a perimeter, it’s a battlefield. From AI-assisted mass exploitation and nation-state pre-positioning, to ransomware hitting hospitals and tribal governments, this week’s episodes of the CyberHub Podcast brought it all into focus. Below is your complete weekly catch-up, organized by threat category with clear action items to take back to your teams.
DATA BREACHES & IDENTITY EXPOSURE
PayPal Credential Stuffing Attack
PayPal confirmed unauthorized access to user accounts after attackers leveraged credential stuffing techniques using reused passwords. Exposed data includes names, addresses, and in some cases tax identification numbers. While PayPal stated no direct financial losses occurred, the downstream risk is significant — fraud operations can mine this identity-linked financial account data for long-term exploitation including synthetic identity fraud and account takeover.
This was not an infrastructure failure. It was a credential hygiene failure, and that distinction matters enormously from a response standpoint.
12.4 Million CarGurus Accounts Exposed
CarGurus disclosed a breach impacting approximately 12.4 million accounts. The exposed dataset includes names, email addresses, hashed passwords, physical addresses, financial pre-qualification data, dealer account details, and subscription information. At this scale, it becomes a credential stuffing gold mine. Even without full financial data confirmation, the combination of identity metadata dramatically increases phishing success rates across other platforms due to password reuse.
Wynn Resorts Employee Data Breach
Wynn Resorts confirmed an employee data breach tied to an extortion attempt. Unlike traditional encryption-based ransomware, this was a data-exfiltration-first model designed to coerce payment. Stolen data reportedly includes employee personal information, potentially Social Security numbers and HR records. Employee datasets are uniquely dangerous — they enable insider impersonation, payroll fraud, and executive-level social engineering.
RANSOMWARE ATTACKS
University of Mississippi Medical Center Shuts Clinics
Ransomware forced the University of Mississippi Medical Center to shut down multiple clinics, disrupting patient scheduling and care continuity. Healthcare ransomware has evolved from financial extortion to operational paralysis — impacting patient safety directly. This is no longer just a cyber problem; it is a patient safety emergency.
Cheyenne and Arapaho Tribes Hit by Ryceta Ransomware
The Ryceta ransomware group targeted tribal government systems, disrupting schools and exposing sensitive citizen data. Attackers demanded 10 Bitcoin. Public sector and tribal entities frequently lack the defensive budgets of private enterprises yet handle highly sensitive governmental and personal data. The risk is twofold: exposure of tribal citizen data and disruption of essential community services.
Advantest Semiconductor Supply Chain Disruption
Semiconductor testing giant Advantest was hit by ransomware, creating potential ripple effects across the chip supply chain. Manufacturing nodes are attractive targets because downtime is expensive and cascading. This isn’t just an IT outage — it’s economic leverage against a globally connected supply chain.
Ransomware Surge in Financial Institutions
New reporting confirms that approximately 65% of financial institutions were hit by ransomware in 2024 — the highest rate across all industries. Recovery costs, excluding ransom payments, are averaging nearly $2.8 million. Threat actors are exploiting legacy systems, weak segmentation, and third-party vendor access, creating both operational and regulatory exposure simultaneously.
“Lock down your crown jewels so that if anything happens, it’s happening on the outskirts with limited scope. You can justify that to the regulator but if you lose your crown jewels, you’ll have a hard time justifying that to your regulator and your boss after the fact.”
NATION-STATE & GEOPOLITICAL THREATS
Volt Typhoon: China’s Long Game in U.S. Critical Infrastructure
Researchers confirmed that China-linked Volt Typhoon remains embedded in U.S. critical infrastructure — targeting energy, water, and telecom environments. This is strategic pre-positioning for geopolitical leverage, not immediate disruption. Living-off-the-land techniques make detection difficult. The message for defenders: assume persistence until proven otherwise through active threat hunting.
Google Disrupts Chinese Espionage Campaign Across 42 Countries
Google announced the disruption of a Chinese espionage operation — attributed to UNC2814 — targeting telecom providers and government entities across 42 countries. Attackers used credential phishing and infrastructure compromise for long-term access inside telecom networks. Most alarming: some backdoors leveraged Google Sheets as a covert command-and-control channel. Telecom compromise grants access to metadata, lawful intercept capabilities, and political communications.
Cyber Operations Guiding Russian Missile Strikes in Ukraine
New intelligence reporting suggests Russian cyber intrusions into Ukrainian infrastructure directly supported kinetic missile strikes — with surveillance access and targeting intelligence improving strike precision. Cyber reconnaissance is becoming operational battlefield intelligence.
“The more Russia uses cyber for kinetic warfare, the more they’re writing the playbook. This playbook will eventually turn from the Russia-Ukraine battlefield to every single place on planet Earth, including our very own, where cyber warfare will lead to kinetic warfare.”
Russian Hybrid Operations Targeting Europe & Romanian Ransomware Links
Dutch authorities report escalating Russian hybrid cyber operations targeting European governments, infrastructure operators, and political institutions. Separately, Romanian officials warn that certain ransomware groups appear strategically aligned with Moscow’s geopolitical objectives — converting criminal ecosystems into destabilization infrastructure. When crime aligns with state interests, attribution and deterrence become exponentially more complex.
Ukraine National Bank Contractor Breached
Attackers compromised a contractor linked to Ukraine’s National Bank, reinforcing that vendor access remains a preferred attack vector. Even when core systems are hardened, contractors become pivot points into financial infrastructure. This is a recurring theme across the week: third parties are the soft underbelly of secure organizations.
AI AS BOTH WEAPON AND TARGET
AI-Assisted Attacks Breach 600 FortiGate Firewalls in 5 Weeks
The headline of the week: an attacker used AI-assisted automation to identify and exploit vulnerabilities across approximately 600 exposed FortiGate firewalls within just five weeks. No sophisticated zero-day required — just automation, scale, and misconfiguration. Exploitation timelines are collapsing. AI dramatically reduces reconnaissance-to-compromise cycles.
“An attacker used AI assistance to breach 600 FortiGate firewalls in five weeks. Mind-blowing, right? Mind-blowing. It’s coming, folks.”
AI Model Distillation Attacks Against Anthropic
Anthropic’s Claude model is facing large-scale model distillation attacks, where adversaries — allegedly including Chinese AI firms like DeepSeek and Moonshot AI — systematically query the system to approximate its decision-making and replicate behavioral patterns. This isn’t classic data theft; it’s intellectual property extraction at scale. Distilled models likely lack the security and ethical safeguards built into Western counterparts.
“China loves to copy, hates to invent. That’s the Chinese model. We let you do all the R&D, you invest all the money, you create, you go first to market, create market demand — then we steal it and offer it for cheaper. That’s the China model. And we’ve allowed it to happen. It’s time to put an end to it.”
AI Vulnerability Tool Impacts Stock Prices
A new AI-based vulnerability analysis platform influenced stock movements after identifying vulnerabilities in publicly traded companies — marking a new frontier where AI-driven security disclosures directly affect capital markets. Organizations must establish coordinated disclosure processes before AI findings go public.
“If AI ruining the stock market or tech stocks doesn’t jolt you, or AI being used to attack 600 FortiGate firewalls exposed to the internet in a matter of weeks doesn’t jolt you a little bit, I don’t know what will. I really don’t know what will.”
UAE Disrupts AI-Assisted Terror Plot
UAE authorities disrupted an AI-assisted extremist operation where AI tools were used to accelerate reconnaissance, automate targeting, and scale propaganda. This confirms what security practitioners have warned: AI misuse is no longer theoretical. Ideological radicalization and AI acceleration are converging — amplifying asymmetric threat capabilities in ways every Western democracy must take seriously.
Claude AI Exploited in Mexican Government Campaign
Researchers uncovered a scenario where Claude AI was abused through persistent prompting to bypass safety guardrails and identify over 20 vulnerabilities across Mexican government systems. This is AI misuse, not an AI breach. When AI-generated outputs are executed without human review, an indirect attack surface is created. AI governance without execution control is theater.
CRITICAL VULNERABILITIES & EXPLOITS
Cisco SD-WAN Zero-Day Exploited Since 2023
Cisco confirmed a critical SD-WAN vulnerability (CVE-2026-21217) had been exploited as a zero-day since 2023 — potentially spanning years of undetected compromise. SD-WAN sits at the network perimeter, granting lateral movement into distributed enterprise and government infrastructure. CISA issued emergency directives for federal agencies. If running SD-WAN, audit logs back to initial exposure windows and conduct retroactive threat hunts immediately.
BeyondTrust PAM Vulnerability Actively Exploited in Ransomware
CVE-2026-1731 in BeyondTrust’s privileged access management platform is now being actively exploited in ransomware campaigns. When remote support infrastructure is compromised, attackers skip reconnaissance and land with privilege. Patch immediately, rotate credentials, and restrict exposed remote support interfaces.
Microsoft Entra Device Code Phishing
Attackers are now abusing legitimate Microsoft device code authentication flows to bypass traditional phishing detection. Victims are tricked into entering legitimate device codes into Microsoft’s own login portal — effectively handing over authentication tokens. No malware, no fake domains. Traditional detection models struggle because the login is technically legitimate. Identity is the new perimeter, and attackers know it.
VMware Aria RCE Vulnerabilities
New vulnerabilities in VMware Aria Operations (CVE-2026-22719 and related CVEs) could allow remote code execution. Aria is used for infrastructure monitoring — meaning compromise doesn’t just provide access, it provides visibility into enterprise workloads. Management interfaces must be restricted to private networks with strict IP allow listing and patched immediately.
SolarWinds Serve-U Critical Flaws
SolarWinds patched four critical vulnerabilities in its Serve-U managed file transfer software enabling potential remote code execution and privilege escalation. Externally exposed FTP systems handling sensitive business data are high-value targets — patch immediately.
Roundcube Webmail Actively Exploited
CISA added newly patched Roundcube webmail vulnerabilities to the Known Exploited Vulnerabilities (KEV) catalog, confirming active exploitation. Attackers are reverse-engineering patches within days, and AI acceleration is only shrinking that window. Patch velocity must match threat velocity.
Grandstream VoIP Critical Vulnerabilities
Critical vulnerabilities in Grandstream VoIP phones could allow attackers to intercept calls or pivot laterally into enterprise networks. VoIP remains one of the most overlooked attack surfaces. Authentication bypass in telephony systems enables both surveillance and internal compromise.
SUPPLY CHAIN, VENDOR RISK & INDUSTRY
Conduent Breach Expands to 25M+ Victims
Conduent’s breach now impacts more than 25 million individuals across multiple states. As a multi-tenant government services provider, eight terabytes of data were reportedly stolen with exposure cascading across clients. This is a vendor concentration risk issue at its most severe: shared infrastructure amplified the blast radius exponentially.
SonicWall Backup Flaw Tied to Ransomware Lawsuit
A lawsuit alleges a breach in SonicWall’s cloud backup solution enabled ransomware fallout — signaling a growing trend: vendor security failures are increasingly triggering legal accountability. Backup infrastructure is resilience insurance; when backup systems become attack vectors, the blast radius becomes catastrophic. Security vendors are no longer immune from downstream liability scrutiny.
Lazarus Group Linked to Medusa Ransomware
North Korea’s Lazarus Group has been linked to Medusa ransomware operations, further blurring the line between financially motivated crime and state-sponsored activity. Lazarus mixes espionage and revenue generation to fund regime activities. When ransomware becomes both criminal enterprise and state funding mechanism, attribution complexity spikes and response strategy must account for geopolitical dimensions.
“When ransomware funds regimes, it stops being just crime — it becomes strategy.”
GitHub Issues Abuse in AI Coding Workflows
Researchers demonstrated how GitHub Issues can be abused to influence AI-assisted coding workflows like Copilot. Malicious prompts embedded in repositories can manipulate maintainers or influence generated code. This blends social engineering with AI development pipelines — making manual code review of AI-assisted pull requests non-negotiable.
FBI Warns of ATM Jackpotting Surge
The FBI flagged an increase in ATM jackpotting attacks where criminals exploit firmware weaknesses or remote management flaws to force machines to dispense cash. Losses already exceed $20 million this year. This is cyber enabling physical crime at the street level — and directly fueling criminal enterprises.
Air Cote d’Ivoire Cyber Incident & Logistics Phishing Campaign
Air Cote d’Ivoire disclosed a cyberattack affecting internal systems, while separately a widespread phishing campaign targeting freight and logistics firms across the U.S. and Europe saw attackers impersonating partners and embedding malware in shipment documentation. Travel and logistics data are highly monetizable — useful for identity fraud, phishing campaigns, and supply chain disruption including real-world shipment rerouting.
U.S. Sanctions Russian Exploit Broker Operation Zero
The U.S. Treasury sanctioned Russian exploit broker ‘Operation Zero,’ accused of buying and selling zero-day vulnerabilities and allegedly offering millions for exclusive vulnerability access. This represents a strategic shift: sanctioning the vulnerability supply chain itself, not just downstream operators.
Chicago Public Schools $17M Breach Settlement
Chicago Public Schools reached a proposed $17 million settlement tied to breach litigation. Breach impact doesn’t end at remediation — it extends into courts, settlements, and long-term institutional trust erosion. The education sector continues facing compounding legal and financial fallout.
Cybersecurity M&A at Record Pace
SecurityWeek reported 426 cybersecurity M&A deals in 2025, totaling nearly $92.5 billion in disclosed value. The market is consolidating around identity, cloud security, and integrated platforms. The question every CISO must answer: are you buying a feature, a product, or a platform — because the market is deciding for you.
White House AI Initiative: Sovereignty Over Global Governance
The White House unveiled a new AI initiative emphasizing American AI infrastructure exports and rejecting centralized global AI governance. AI is now declared economic and national security infrastructure. Expect AI compliance fragmentation globally — if data privacy was complex, AI governance will be exponentially more so.
China Domestic IP Crackdown: Strategic Posturing?
China announced a domestic intellectual property crackdown. While appearing to be internal reform, this aligns strategically with geopolitical positioning amid global criticism of Chinese cyber espionage. Multinational organizations operating in China must maintain strict data segregation and governance separation. Trust nothing by default. Validate everything by design.
Robot Vacuum IoT Surveillance Nightmare
An engineer reverse-engineering a robot vacuum ecosystem discovered an unauthenticated API exposing control over nearly 7,000 devices across 24 countries — including live camera feeds, microphone audio, and mapping data. IoT devices are no longer novelty gadgets; they are distributed sensors embedded in homes and offices globally. Weak backend controls can turn consumer devices into surveillance networks.
THIS WEEK’S ACTION LIST
Take these prioritized steps back to your teams this week:
1. Enforce MFA with device binding on all financial and high-risk platforms — reset credentials for any accounts involved in the PayPal or CarGurus breaches
2. Conduct retrospective forensic audits on Cisco SD-WAN deployments going back to 2023 exposure window
3. Patch BeyondTrust, SolarWinds Serve-U, VMware Aria, Roundcube, and Grandstream VoIP immediately — do not delay
4. Remove public-facing firewall and management interfaces; restrict to private networks with IP allow lists
5. Implement continuous vendor and third-party access monitoring with time-bound credentials — no persistent vendor access by default
6. Deploy network micro-segmentation in healthcare environments and segment VoIP, IoT, and OT assets from enterprise networks
7. Conduct proactive OT threat hunting for Volt Typhoon-style living-off-the-land persistence techniques
8. Require manual code review for all AI-assisted pull requests and restrict AI-generated outputs from automated execution without human review
9. Implement anomaly detection on AI API query volume to identify model distillation reconnaissance
10. Tightly restrict device code authentication flows in Microsoft Entra environments where operationally unnecessary
11. Conduct external ransomware readiness audits in healthcare, public sector, and tribal government environments
12. Integrate geopolitical threat modeling into ransomware response plans — attribution now carries strategic implications
13. Establish coordinated disclosure guardrails before AI-driven vulnerability findings reach public channels
14. Demand tenant-level logical data segregation from multi-tenant vendors; validate backup integrity independently from vendor assurance
JAMES AZAR’S CISO’S TAKE
This week’s episodes reinforced a single, unavoidable conclusion: convergence is the defining threat of our era. AI extraction, AI-assisted terror plots, cyber-enabled missile strikes, ransomware aligned with geopolitical interests, nation-state pre-positioning in critical infrastructure, and automated mass exploitation, these are no longer isolated incidents. They are connected threads in a coordinated strategic tapestry. The offensive cycle is accelerating faster than most organizations are prepared to handle, and the old model of separating cybercrime, espionage, and geopolitics has collapsed. It is all interconnected now, and every CISO who hasn’t updated their threat model to reflect that reality is operating with an obsolete map.
From my perspective, the response isn’t panic it’s precision. This week’s stories demand that we obsess over crown jewels, segment aggressively, assume persistence until proven otherwise, treat AI workflows as attack surfaces requiring governance, and elevate geopolitical threat modeling into board-level discussions. The perimeter isn’t gone it’s multiplied. Every management interface, every AI prompt workflow, every SD-WAN edge device, every cloud backup is perimeter now. Our role as CISOs has expanded beyond security operations; we are now custodians of economic stability, institutional trust, and national infrastructure. Patch faster. Segment deeper. Validate continuously. The adversary isn’t waiting and neither should we.
Stay Caffeinated. Stay Vigilant. Stay Cyber Safe.