Cyber Attacks & Data Breaches: September 2026
A month-end digest of ransomware, data breaches, zero-days, supply chain compromises, gang wars and a dedicated look at AI-driven attacks.
Read this first: this page is compiled from public reports and trackers. Counts differ between sources (leak-site claims vs confirmed attacks), attributions are often preliminary, and claimed breaches may be unverified. Always verify with the original advisory before acting.
🔒Ransomware & Extortion
Ransomware cooled from August’s record but stayed well above 2025 levels. Confirmed attacks were far fewer than leak-site claims, and Qilin led confirmed incidents.
Nutex Health
A ransomware gang claimed it stole company and patient data. The company had not confirmed the claims or scope.
ClaimedJack Henry & Associates
The banking-technology provider was hit by ransomware.
RansomwareVeradigm
The Gentlemen listed the healthcare-tech firm on its leak site, claiming 3.5M records. The incident was confirmed, the figure was not.
HealthcareNEPSE (Nepal Stock Exchange)
A “technical glitch” was a ransomware attack on a data center provider. Trading was suspended for a day and systems restored from backups.
FinanceKeio Corporation
The Japanese railway group confirmed ransomware disrupted business systems.
TransportFosko S.A. (Chile)
Ransomware hit production computers and shared files at a plastic packaging maker.
ManufacturingSpringfield Public Schools, MA
A major ransomware attack closed schools for several days and disrupted email, computers and student medical records.
EducationRohloff Group (KFC franchisee, Africa)
INC Ransom claims 536GB across 103,000+ files, mostly employee data.
ClaimedSettra ransomware
A newly reported variant seen in retail and manufacturing attacks, using RMM tools, recovery disruption and BYOVD techniques.
New strain🗄️Data Breaches & Leaks
Identity data drove the biggest numbers, alongside large healthcare exposures and ShinyHunters extortion.
IDScan.net
The ID-verification provider confirmed unauthorized access. Reports linked it to a dark web service advertising 153M+ US and Canadian driver’s license scans.
IdentityAesto Health
Over 9.5 million patient records affected after a third party reached data through an external-facing system.
HealthcareMcKesson
ShinyHunters exposed data on 6.4 million people in an attack on the medical supplier.
HealthcarePentagon Defense Manpower Data Center
Over 3 million people affected; unencrypted personnel data was reachable through a file-sharing flaw.
GovernmentMcDonald’s Indonesia
About 2.8 million customer records exposed in a data leak.
RetailGyazo (Helpfeel)
An upload-server vulnerability gave an attacker database access, affecting about 23.62 million user records.
Web serviceCondé Nast (alleged)
A database said to hold 32.8 million user records was offered for sale for $15,000 on a cybercrime forum.
Alleged saleMedela
ShinyHunters claimed the breach; leaked data exposed 423,947 accounts, mainly healthcare professionals.
HealthcareRingCentral (claimed)
ShinyHunters claimed 623GB stolen after a social engineering attack.
ClaimedAstrana Health
Attackers impersonated staff and spoofed the main phone number to trick employees into giving access.
Social engineeringSIMBA (Singapore)
A breach affected 23,549 telecom customers and is under regulator investigation.
TelecomTrezor via ShipMonk
Confirmed exposure of about 81,000 customers through a logistics partner.
Third party⚡Zero-Days, Exploits & Edge-Device Penetration
A record Patch Tuesday, exploited email and network gateways, and ransomware gangs targeting CI/CD servers.
SonicWall SMA1000
SSRF and OS command injection flaws (CVE-2026-83548, -83549) were listed among actively exploited vulnerabilities.
ExploitedMicrosoft Patch Tuesday
A record 974 CVEs fixed, including two exploited privilege-escalation zero-days (CVE-2026-85880, CVE-2026-81963) and 20 potentially wormable bugs.
Zero-dayCisco Secure Email Gateway
CVE-2026-76461 let unauthenticated attackers gain root through a crafted email; CISA added it to KEV.
Zero-dayGoogle Pixel modem
CVE-2026-58704 allowed privilege escalation with no user interaction; CISA gave agencies three days to patch.
Zero-dayJetBrains TeamCity
Ransomware gangs exploited CVE-2026-63077 to bypass authentication and run OS commands, risking credentials and CI/CD pipelines.
Auth bypass🔗Supply Chain & Malware Campaigns
Trusted vendors and developer ecosystems stayed prime targets.
Brevo supply-chain attack
Compromised Cloudflare access was used to inject ClickFix scripts, potentially affecting over 100,000 websites.
Supply chainTanStack leak follow-up
CrowdSec’s final report tied a September source-code leak to the May TanStack compromise by TeamPCP.
Supply chainNodeRabbit RAT
A new Node.js RAT linked to Iranian Nimbus Manticore activity, delivered through fake coding-test archives.
Malware🌐Hacktivism, Gang Wars & State-Linked Activity
ShinyHunters hijack Clop’s leak site
Using a Grav CMS flaw, the group defaced Clop’s leak site and claimed to steal server data and onion-service keys.
Gang vs gangFBI jobs portal (claimed)
ShinyHunters claims 2–3TB of FBI personnel and applicant data. The FBI is investigating and the claim is unverified.
ClaimedATNS (South Africa)
Ransomware-linked malware was reportedly found in air-traffic OT networks, with data sent to IPs in China.
Critical infrastructureWest Bengal government repository
A major cyberattack crippled a state data repository, with no stolen information recovered.
Government🤖AI-Driven Cyber Attacks
September showed AI agents compressing intrusions from weeks to hours. Google notes most cases are agent-enabled with humans still steering, not fully autonomous.
First fully documented agentic ransomware case
Palo Alto Networks’ Unit 42 said a hacker used frontier AI models to breach a European software company in under 10 hours, work that normally takes about two weeks.
Agentic ransomwareAI agents vs government systems
Hunt.io reported Chinese-speaking hackers using Claude, Qwen and DeepSeek agents against government targets, noting such workflows amplify flawed assumptions too.
Agentic reconChinese hackers use AI on stolen networks
Google found AI-powered operations, some under six hours end to end, with Mandiant’s Mandia describing “hacking 24–7 with total recall”.
State-linkedGambit Security: 100-target AI campaign
A Chinese-speaking financially motivated operator used open-source AI agents against up to 100 organizations in five days, stealing 600K+ card records.
Agentic fraudGoogle AI Threat Tracker
A suspected criminal ran a multi-agent framework that harvested thousands of credentials in under six hours. Google calls it agent-enabled, not fully autonomous.
Credential theftSpain’s first AI-agent breach notification
The AEPD received its first personal-data breach notification involving an attack run through an AI agent; the investigation continues.
Regulatory first- Speed: breaches that took two weeks now take under ten hours.
- Scale: one operator can target around 100 organizations in days.
- Defense: automate detection and response, review AI tool permissions, and patch internet-facing gateways first.
- CM-Alliance, SOCRadar, Intellizence, Allen DeVaux, BlackFog and CSIS incident roundups
- Ransomnews, Breachsense and rSecurity (ransomware counts); SecurityWeek, Security Affairs and eSecurityPlanet (vulnerabilities, supply chain)
- Semafor, Bitdefender, GBHackers, SecNews, IBTimes and iHeart (AI-driven attacks)
Independent roundup. Not affiliated with any vendor named. Next edition: October 2026.
