Daily Summary of Cybersecurity · Final editionTuesday 17 November 2026 · UTC
Cybersecurity News: 17 November 2026
Java4Shell hits OpenJDK as TSMC breach and NSA Net disruption follow
Researchers disclosed Java4Shell, a critical OpenJDK Base64 flaw that can give unauthenticated attackers code execution in major Java applications, while Oracle shipped a fix. TSMC confirmed employee data was exposed through an unnamed cloud provider after a cybercriminal forum post. The NSA said its classified NSA Net suffered temporary degradation during routine maintenance, adding to a week of high-profile outages.
Critical Java4Shell flaw hits OpenJDK and major apps
Calif Research disclosed Java4Shell, CVE-2026-114938, a memory-corruption flaw in OpenJDK's Base64 decoder that can let unauthenticated attackers run code in Keycloak, Jira, Confluence, Jenkins and PingFederate. The bug affects OpenJDK 17.0.6 and later on AVX-512 x86-64 systems, and Oracle shipped a fix the same day it released its November update. Calif rated it 9.2 critical, while Oracle assigned 8.1 high.
NSA says NSA Net degradation was routine maintenance
The NSA said its classified internal network, NSA Net, suffered temporary degradation during scheduled routine maintenance, with mission-critical capabilities remaining operational. The agency said its infrastructure has multiple layers of redundancy, and people familiar with the matter said the issue was not believed to be a cyberattack. The disruption added to a week of high-profile outages, though no evidence links them.
Oracle to patch 11 Java flaws as OpenJDK RCE report emerges
Oracle will patch 11 Java SE vulnerabilities on Tuesday, all of which it says may be exploited remotely without authentication and have a maximum CVSS score of 8.1. A separate report warns of a potentially critical OpenJDK remote code execution flaw, CVE-2026-114938, but details remain unconfirmed and it is not yet known to be among the 11. Oracle advises customers to stop using current builds of Java 8, 11, 17, 21, 25 and 27 after the update.
TSMC confirms employee data breach linked to cloud provider
TSMC said personal data of some current and former employees was exposed after a breach at an unnamed cloud provider, while its own systems and customer data were unaffected. The matter surfaced after a November 8 forum post offered 6.9 GB of alleged employee data for 3 Bitcoin, about $200,000. TSMC verified the records matched the provider's systems and is notifying affected staff while working with the vendor and authorities.
NIST has published a roadmap for moving to post-quantum cryptography, with deadlines for risk mitigation measures set for 31 December 2030. A GAO report says US federal agencies are not yet prepared to meet them, leaving encrypted data exposed to harvest now, decrypt later risks. The guidance also applies to technology providers, standards organizations and infrastructure operations on which federal systems depend.
Ipsos investigating cybersecurity incident after filing complaint
Market research firm Ipsos said on Monday that it is investigating a cybersecurity incident affecting one of its internal information systems. The company filed a complaint on November 11, activated incident response, and brought in outside security experts, but has not said whether data was accessed or stolen. Ipsos said the incident has not disrupted operations or client delivery and has reported the matter to authorities.
New GitDoor and GraphDoor Backdoors Turn GitHub and SharePoint Into Covert C2 Channels
Kaspersky said Operation ShadowCloud used a MeshCentral Agent to install GraphDoor and GitDoor backdoors in government sectors in Asia and Africa. The backdoors use SharePoint, GitHub and Ably as covert command-and-control channels and can download plugins or run shellcode in memory. Kaspersky traced the start to an unspecified Taiwanese organization and said the infection may have reused a prior compromise.
Visa and JPMorgan face ransomware claims as n0n sets 48 hour leak deadline
A newly emerged ransomware gang called n0n has listed Visa and JPMorgan Chase as alleged victims and given them 48 hours to negotiate before leaking stolen data. Visa said the incident began with a third party service provider in the Asia Pacific region and that its systems and card data were not impacted. JPMorgan Chase said a software bug caused a service outage and saw no evidence of a cyberattack. The claims target two major financial firms.
Court clears brothers of $25 million Ethereum crypto heist charges
US District Judge Jessica Clarke acquitted former MIT students Anton and James Peraire-Bueno of wire fraud and related charges over a scheme that stole about $25 million in cryptocurrency in about 12 seconds by exploiting an Ethereum blockchain bug. The decision came one year after a mistrial, with the judge finding the government failed to prove material misrepresentations. The case was described as first-of-its-kind.
Microsoft 365 outage fuels credential phishing campaign
Microsoft says attackers are using fake M365 outage notifications to lure users into clicking links that may steal credentials. The emails, observed during a 22-hour outage that began November 12, claim held messages will be deleted within 24 hours unless recipients restore them. Microsoft identified domains including moments-notice.com, planetmaison.com and newbooker.com, and said credential capture and malware delivery are not yet confirmed.
Apple splits Mac app permissions to limit AI agent access
Apple introduced more granular permissions in macOS 27.3 beta 2, splitting Device Control and Data Access into System & App Protection Override and Accessibility. The changes add transparency and let users require apps to ask each time they need access, with a new indicator showing when an app controls another. Existing permissions remain valid, and some developers can request entitlements to preserve prior behavior.
Google says AI agents ran exploits with no human in the loop
Google's GTIG and Mandiant said threat actors used AI frameworks to automate vulnerability discovery and exploitation without a human in the loop for the first time. An exposed staging server ran exploit_agent, which compared application code before and after patches, drafted exploits and ran them. The FBI warned that ARTEX had been used since at least August 2026 against financial services, government, technology, defense, healthcare and energy.
Uber notifies 149,346 drivers after Fireworks breach
Uber is notifying 149,346 drivers that their names, driver's license numbers, dates of birth and addresses were stolen after attackers used compromised credentials at AI infrastructure provider Fireworks to access Uber data. Uber said the provider alerted it on October 9 and that it cut data connections and investigated. The breach follows an October 24 Fireworks incident linked to a financially motivated group.
NATO AI cyber defense work hits up to 300 billion OpenAI tokens a month
NATO's Communications and Information Agency gives NATO core institutions access to OpenAI cybersecurity models to help protect Allied networks, and NATO has used up to 300 billion raw input and output tokens a month through Daybreak. ENISA used the models to find four vulnerabilities in software used across EU institutions, with fixes released. CERT Polska used them to find six vulnerabilities in MikroTik RouterOS.
Greiz Hospital Reopens Emergency Room After Extortion Attack
The Greiz-Ronneburg district hospital in Thuringia reopened its emergency room gradually on Monday evening after a system failure reported Saturday. District spokesman Uwe Müller called the hacking attack an extortion attempt, saying a message reached the hospital on Sunday. Patients continued to receive care during the outage, with emergency cases directed by the Gera control center, though digital treatment plan problems remained.
Kazakh ransomware suspect extradited to US faces up to 27 years
Roman Khlynovskiy, a Kazakh national, was extradited from Italy to the US on charges that he and the 8-Digits Team targeted schools, hospitals and health entities in at least six states. They allegedly stole over 70 million health records and demanded over $500 million in ransoms. Khlynovskiy faces up to 27 years if convicted of computer fraud, identity theft, wire fraud, extortion, money laundering and non-consensual intimate image threats.
Anthropic cuts disclosure deadline for serious vulnerabilities to 30 days
Anthropic will publicly disclose high- and critical-severity vulnerabilities it reports to vendors and open source maintainers after 30 days, down from 90, with all other bugs at 45 days. The change applies to reports sent on or after January 1, 2027, while open reports keep their original deadlines. Anthropic said AI models make rediscovery more likely, and it will escalate unanswered reports after 15 days.
ConnectWise says ScreenConnect attacks continue as SCKit targets unpatched systems
ConnectWise completed its investigation into unauthorized access to partner ScreenConnect environments and advised customers to run technician client 26.6.5 or later. It said activity continues in on-premises systems where identified extensions remain enabled and ScreenConnect is not updated. Version 26.6.5 addresses critical CVE-2026-84869. Huntress reported a November 13 spike in SCKit activity affecting 205 hosts across 41 organizations.
Microsoft adds one Intune policy for Windows, .NET and boot recovery updates
Microsoft expanded Windows Autopatch so administrators can manage Windows quality updates, supported .NET Framework updates, and Quick machine recovery updates through one policy in Microsoft Intune. The controls allow automatic or manual approval, with a recommended automatic deferral of 0 to 30 days for security updates. Quick machine recovery helps Windows devices recover when a large-scale problem prevents them from starting.
Japan's Fugaku supercomputer taken offline after status page hack
RIKEN said an unknown entity attacked a Fugaku status dashboard between November 10 and 13, reached servers holding user names, emails and credentials, and installed malware. The organization took Fugaku offline after an attempted login on November 13, disrupting scheduled research. It says 4,222 Fugaku users and 375 Rikyu users are at risk, though it has found no evidence that research data leaked.
N-able patches 17 N-central flaws as attackers deploy crypto miners
N-able released N-central 2026.4 Hotfix 3 to patch 17 critical flaws in its RMM and UEM platform. The update removes legacy API support and blocks remote control and Intune integration, while customers must upgrade Agents and Probes to avoid impersonation. The release follows 4EverAble2Shell and Agent2Shell bugs, with 4EverAble2Shell flaws exploited to deploy cryptocurrency miners on vulnerable servers.
Optimizely supply chain attack loads malware on Zoom, CNN, BBC and Salesforce sites
Attackers exploited a vulnerability in Optimizely's Experimentation account signup flow to inject malicious JavaScript into customer projects, causing sites for Zoom, CNN, BBC, Salesforce and GitLab to load scripts that can steal cryptocurrency and payment data. Optimizely said it removed the injected code, revoked tokens and reset access for affected accounts. Salesforce said Optimizely scripts are disabled on some websites as a precaution.