> ## Content Index
> Fetch the complete content index at: https://www.cyberverso.net/llms.txt
> Use this file to discover other available public pages before exploring further.

# Cyber / Brief, 18 Nov 2026: Netherlands calls US bid to extend chip controls to ASML unacceptable
- URL: https://www.cyberverso.net/brief/cyber-brief-18-nov-2026/
- Published: 2026-11-18T07:41:35.000Z
- Updated: 2026-11-18T07:41:35.000Z
- Description: The Netherlands has told Washington that the MATCH Act, a bipartisan bill that would extend US controls on chipmaking equipment to ASML's machines unless allies match them within 150 days, is "unacceptable between partners", turning the Nvidia scandal into a question of European…
- Author: Paolo De Rosa
- Tags: #bulletin, #signal, #en

The Netherlands has told Washington that the MATCH Act, a bipartisan bill that would extend US controls on chipmaking equipment to ASML's machines unless allies match them within 150 days, is "unacceptable between partners", turning the Nvidia scandal into a question of European sovereignty over export policy. The same pressure is reshaping the market on both sides: Nvidia says it has stopped shipping the Hopper chips it was licensed to sell to China, Beijing threatens retaliation against the planned remote-access curbs on the eve of the Trump–Xi meeting, and banks are reconsidering billions in loans to Southeast Asian data centres built for Chinese tenants. Malaysia's regulator is pressing Anthropic for details of a Claude-run election-manipulation operation and drafting an AI law, while the FBI warns that the ARTEX framework lets criminals reach privileged access in under three hours and Google shows Russia's Cold Relic chaining one victim's stolen cloud data into the next intrusion. Japan has shut its Fugaku supercomputer after intruders stole users' keys and research files, and Norway's Equinor, a major gas supplier to Europe, has confirmed a cyberattack on its IT systems.

## Top Stories

- [A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI](https://cyberscoop.com/java4shell-critical-java-flaw-openjdk/) — *CyberScoop* · Cybersecurity & Threats
- [Dutch resist ‘unacceptable’ US bid to tighten semiconductor export controls](https://www.politico.eu/article/dutch-resist-unacceptable-us-bid-to-tighten-semiconductor-export-controls-asml/?utm%5Fsource=RSS%5FFeed&utm%5Fmedium=RSS&utm%5Fcampaign=RSS%5FSyndication) — *Technology – POLITICO* · EU & Technology
- [Malaysia Seeks Details From Anthropic Over AI Influence Campaign](https://www.bloomberg.com/news/articles/2026-11-18/malaysia-seeks-details-from-anthropic-over-ai-influence-campaign) — *Bloomberg Technology* · AI & Power
- [Nvidia Stops Selling AI Chips to China Despite US Approval](https://www.bloomberg.com/news/articles/2026-11-18/interim-nvidia-ceo-says-it-s-no-longer-sending-ai-chips-to-china) — *Bloomberg Technology* · China & Technology
- [Japan takes Fugaku supercomputer offlline after cyberattack](https://www.bleepingcomputer.com/news/security/japan-takes-fugaku-supercomputer-offlline-after-cyberattack/) — *BleepingComputer* · Cybersecurity & Threats

---

## AI & Power

[Malaysia Seeks Details From Anthropic Over AI Influence Campaign](https://www.bloomberg.com/news/articles/2026-11-18/malaysia-seeks-details-from-anthropic-over-ai-influence-campaign) — *Bloomberg Technology*  
Why it matters: A government asks a frontier lab for the evidence behind an AI-run election operation, and drafts an AI law.  
Malaysia's Communications and Multimedia Commission has asked Anthropic for more information on its September report, which said Claude had been used to run a commercial election-manipulation platform aimed mainly at Malaysian voters, with about a thousand fake X accounts, a fake news outlet and fabricated dossiers, Bloomberg reports. The communications ministry told parliament the regulator will then decide whether legal action is needed, and that a bill to regulate AI use is being drafted.

[Anthropic Warns AI Safety Tests Are Becoming Too Easy](https://www.healthcareinfosecurity.com/anthropic-warns-ai-safety-tests-are-becoming-too-easy-a-33424) — *HealthcareInfoSecurity.com RSS Syndication*  
Why it matters: A lab says single-prompt safety tests flatter models, as evaluations become the basis of regulation.  
Anthropic says safety evaluations built on single harmful requests are saturated: models refuse them easily, while real misuse escalates gradually, so a model that handles the first message well may fail by the fifth. It calls for multi-turn tests, difficulty tuned to current capabilities, sandboxed runs and results broken down by harm category rather than a single headline refusal rate.

[Exclusive: Donald Trump Jr. signals investors underestimated AI pushback](https://www.axios.com/2026/11/17/donald-trump-jr-data-centers) — *Axios*  
Why it matters: Inside the Trump circle, an admission that the AI industry is losing the public argument.  
Donald Trump Jr. told Axios that investors underestimated the political pushback against AI, after midterms in which voters approved dozens of local moratoriums on data centres. He said much of the backlash is "manufactured" but still real, and that the founders of the biggest AI companies have not served themselves well as their own marketing department. Excerpt-based.

[US-China AI race causing global ‘alarm,’ UN's top tech envoy says](https://www.semafor.com/article/11/17/2026/us-china-ai-race-causing-global-alarm-uns-top-tech-envoy-says) — *Semafor*  
Why it matters: The UN's tech envoy says the two-power race is making the rest of the world fear for the rules.  
Amandeep Singh Gill, the UN Secretary-General's envoy on digital technologies, told Semafor that other countries watch the US–China AI race "with some kind of alarm": they want their own models and infrastructure, and fear that racing means ignoring guardrails after recent safety incidents suggested "shortcuts are being taken". He called for a large infusion of compute, connectivity and energy for Africa.

---

## EU & Technology

[Dutch resist ‘unacceptable’ US bid to tighten semiconductor export controls](https://www.politico.eu/article/dutch-resist-unacceptable-us-bid-to-tighten-semiconductor-export-controls-asml/?utm%5Fsource=RSS%5FFeed&utm%5Fmedium=RSS&utm%5Fcampaign=RSS%5FSyndication) — *Technology – POLITICO*  
Why it matters: The Hague calls a US bill that would extend American controls to ASML's machines an attack on its sovereignty.  
The Dutch foreign ministry told POLITICO that the extraterritorial reach of the bipartisan MATCH Act is "unacceptable between partners". The bill would give allies such as the Netherlands and Japan 150 days to match US limits on selling and servicing chipmaking equipment in China, after which Commerce would have to apply US controls to their equipment unilaterally. It has gained momentum since the revelations about Jensen Huang and a suspected Chinese spy. China was nearly a third of ASML's revenue last year, and the Dutch already restrict its exports but have resisted going further.

[Europe the natural location for Mistral IPO, CEO says](https://www.semafor.com/article/11/17/2026/europe-the-natural-location-for-mistral-ipo-ceo-says) — *Semafor*  
Why it matters: Europe's sovereign-AI champion says it will list at home, a test of whether EU capital markets can hold one.  
Mistral CEO Arthur Mensch said the company is preparing to go public and that "the natural place to list is Europe", though not this year; most of its investors and half its customers are European, and it pitches itself as a sovereign alternative to closed US and open Chinese models. Arm, Spotify and Klarna chose New York for deeper capital. Mensch also tied France's youth unrest to anxiety about AI and jobs.

[Investment in European startups surges to $81BN, but Europe faces challenges, says Atomico](https://tech.eu/2026/11/18/investment-in-european-startups-surges-to-81bn-but-capital-concentration-is-a-challenge-for-startups-says-atomico/) — *Tech.eu*  
Why it matters: Record European founding and funding, held back by compute, exits and concentrated capital.  
Atomico's State of European Tech report estimates a record 40,000 new tech companies founded in Europe in 2026 and $81 billion invested, the second-highest ever, with the ecosystem valued at $4.4 trillion and over 450 companies worth more than $1 billion. Funding is concentrated in fewer, larger rounds: pre-Series A rounds fell almost 30% between 2021 and 2025\. The report names the lack of AI compute and weak exits as the main constraints.

[Europe needs ‘innovation-friendly’ regulation to catch up on AI, German official says](https://www.semafor.com/article/11/17/2026/europe-needs-innovation-friendly-regulation-to-catch-up-on-ai-german-official-says) — *Semafor*  
Why it matters: Germany's digital ministry says the AI Act hurts European firms more than American ones.  
Marco-Alexander Breit, director-general for digital policy at Germany's digital ministry, told Semafor that Europe needs a more "innovation-friendly" regulatory framework in Brussels to catch up on AI, and that Germany hopes to quadruple its data-centre capacity by 2030\. He said he agrees with most of the AI Act but that it raises the threshold for European companies and "hurts our economy" more than US companies.

---

## US & Technology

[Nvidia says it does not intend to reinstate Jensen Huang as CEO](https://www.theverge.com/report/1035465/nvidia-says-it-will-not-reinstate-jensen-huang-as-ceo) — *The Verge*  
Why it matters: Nvidia's board shuts the door on Huang despite pressure from Trump and Silicon Valley.  
Opening Nvidia's earnings call, new board chair Harvey Jones said the board unanimously accepted Jensen Huang's resignation and "does not intend to reinstate" him, and took no questions on the circumstances or the ongoing investigation. Huang had said hours earlier that he would return if asked; Trump said he "would not mind". Huang is due to testify before Congress on 1 December, and Democrats taking control of Congress plan to investigate the administration's dealings with Nvidia.

[Clampdown on China Chip Loophole Puts Billions at Risk](https://www.bloomberg.com/news/articles/2026-11-17/clampdown-on-china-chip-loophole-puts-billions-in-loans-at-risk) — *Bloomberg Technology*  
Why it matters: The new licence rule for Southeast Asian data centres reaches the banks that financed them.  
Lenders are reviewing data-centre loans in Southeast Asia after Monday's US rule requiring operators in Singapore, Malaysia, Indonesia and Thailand to obtain licences to buy advanced US chips, Bloomberg reports. Asia-Pacific operators have borrowed about $49 billion since 2023, often secured on contracts with Chinese tenants and sometimes on the chips themselves. Five lenders are discussing leaving a $480 million loan to Aolani Cloud for chips also used by ByteDance, and a $300 million loan to GMI Cloud for a Thai site serving Tencent would have to be repaid if Tencent were restricted.

[Rubio Says Recent US Tech Outages Unrelated, Not Hostile Acts](https://www.bloomberg.com/news/articles/2026-11-18/rubio-says-recent-us-tech-outages-unrelated-not-hostile-acts) — *Bloomberg Technology*  
Why it matters: Washington says the week of outages was not an attack, without saying what caused each one.  
Secretary of State Marco Rubio said in Manila that the US has no indication the recent outages, including the nationwide air-traffic ground stop and the Microsoft disruption, are related or the work of hostile actors: "We are not under systemic attack." He said each has an explanation unrelated to criminals or another country, but would not say whether AI was involved, and acknowledged the series is unsettling. Apple's sign-in outage on Tuesday, which it blamed on an internal update, was the latest.

[How CISA is prioritizing companies for talks about security improvements](https://cyberscoop.com/cisa-100-critical-infrastructure-companies-fed-talks/) — *CyberScoop*  
Why it matters: CISA picks about 100 infrastructure owners where adversaries have pre-positioned, for resilience talks.  
Acting CISA director Nick Andersen said the agency has identified about 100 critical-infrastructure companies for a first round of talks on hardening their systems, chosen by crossing a homeland-defence working group's list of the most vital functions with five to six years of intelligence on where adversaries pre-position and invest. The talks will extend to each company's energy, telecoms and IT providers under the "CISA 2030" strategy.

[‘The Predators Are Real’: Senate Hearing Slams Roblox Over Child Sex Abuse](https://www.wired.com/story/the-predators-are-real-senate-hearing-slams-roblox-over-child-sex-abuse/) — *WIRED*  
Why it matters: A bipartisan bill would let families take Roblox to court instead of arbitration.  
People who say they were groomed on Roblox testified before a Senate subcommittee on a bill from Josh Hawley and Dick Durbin that would let victims' families sue the platform instead of being bound by forced arbitration in its terms of service. Roblox faces nearly 200 such lawsuits and its CEO declined to testify; the company says its age checks, protective defaults and chat limits show its commitment to safety, and that 35% of its age-checked daily users are under 13.

---

## China & Technology

[Nvidia Stops Selling AI Chips to China Despite US Approval](https://www.bloomberg.com/news/articles/2026-11-18/interim-nvidia-ceo-says-it-s-no-longer-sending-ai-chips-to-china) — *Bloomberg Technology*  
Why it matters: Nvidia stops shipping its licensed China chips on its own, after selling $2.4 billion of them last quarter.  
Interim CEO Colette Kress told analysts that Nvidia is no longer shipping the licensed Hopper data-centre products it was allowed to sell to China, and that its guidance includes none of them, Bloomberg reports. It sold $2.4 billion of them in China in the third quarter. The halt comes after the Huang scandal put Nvidia's China sales under board supervision, and ahead of this week's Trump–Xi meeting.

[China threatens retaliation against US chip curbs](https://www.semafor.com/article/11/17/2026/china-threatens-retaliation-against-us-chip-curbs) — *Semafor*  
Why it matters: Beijing threatens to answer the US remote-access curbs on the eve of the summit.  
China threatened retaliation if the US goes ahead with plans to restrict remote access to advanced chips, Semafor reports, raising the risk of a new round of tit-for-tat curbs before this week's leaders' summit. An analyst it cites says the remote-access clampdown and harsher export controls would make catching up "a lot more expensive" for Chinese AI startups.

[Nvidia Introduced Alibaba to Alice Huang for Workaround on U.S. Export Controls](https://www.theinformation.com/articles/nvidia-introduced-alibaba-alice-huang-workaround-u-s-export-controls) — *The Information*  
Why it matters: Nvidia's own sales team is reported to have opened Alibaba's door to Alice Huang.  
Citing three people with direct knowledge, The Information reports that Nvidia's sales team introduced Alibaba to Alice Huang twice in 2022, months before US export controls took effect, and that this helped her company win Alibaba as a customer and then obtain bank loans to buy Nvidia chips. Excerpt-based.

[How a Tsinghua spin-out became China’s first commercial AI safety watchdog](https://www.scmp.com/tech/tech-trends/article/3374829/how-tsinghua-spin-out-became-chinas-first-commercial-ai-safety-watchdog) — *Tech - South China Morning Post*  
Why it matters: China gets its first independent frontier-AI evaluator, as AI safety reaches the Trump–Xi agenda.  
Fangcun AI, spun out of Tsinghua University in January, is China's first company dedicated to independent, third-party evaluation and research on frontier AI safety, SCMP reports; it has over 40 staff, recruits from Tsinghua's Yao Class and raised seed and angel rounds from investors including the Tsinghua Alumni Seed Fund. CEO Luke Li Xiaojian said the field had been dominated by US firms. Seed and angel deals rose to 29% of AI-security financing in the first three quarters, according to CVSource.

[China’s ‘token export’ ambitions take shape with new Huawei-powered data centre](https://www.scmp.com/tech/big-tech/article/3374791/chinas-token-export-ambitions-take-shape-new-huawei-powered-data-centre) — *Tech - South China Morning Post*  
Why it matters: China answers the compute curbs by exporting tokens from Huawei chips, not chips.  
A 1 billion yuan cluster of Huawei Ascend 950DT chips at China Telecom's submarine-cable landing station in Shantou started operating last week to serve foreign customers, processing their AI queries locally and sending only the generated tokens back, according to local media cited by SCMP. Chinese models from MiniMax, Moonshot, Z.ai, StepFun, DeepSeek and Xiaomi have ranked near the top of global token usage this year.

[US President Trump casts doubt on spy claim at heart of Nvidia saga](https://www.scmp.com/tech/big-tech/article/3374849/us-president-trump-casts-doubt-spy-claim-heart-nvidia-saga) — *Tech - South China Morning Post*  
Why it matters: The White House is split on whether the woman at the centre of the Nvidia case is a spy.  
Trump said he had not heard intelligence that Alice Huang was a Chinese operative, contradicting reports that Marco Rubio had briefed him, and suggested such claims are routinely made about women from China, SCMP reports. Treasury Secretary Bessent called the spy question a "canard" and said the issue was export-control evasion, while the attorney general said the Justice Department had reached no conclusion. Beijing calls the claims a smear.

---

## Threat Intelligence (CTI)

**\[P2\]** [Russian COLD RELIC Hackers Use Stolen Google Workspace Data to Breach Second U.S. Organization](https://thehackernews.com/2026/11/russian-cold-relic-hackers-use-stolen.html) — *The Hacker News*  
Why it matters: Russian intelligence hackers chain one victim's cloud data into the next intrusion.  
Google's Mandiant and GTIG say Cold Relic (COLDRIVER, Star Blizzard) used the staff directory, data and identities stolen in a late-2025 compromise of one US organisation's Google Workspace to breach a second in 2026\. In the first, it used a Super Admin's stolen credentials and MFA fatigue through a nearby residential proxy, registered its own device, passkeys and backup codes, downloaded the whole cloud drive and deleted Alert Center notifications.  
severity high · exploited in the wild · EU: NIS2 · actor Cold Relic (COLDRIVER) (80%)

**\[P2\]** [FBI Says ARTEX AI Tool Helped Hackers Gain Privileged Access in Under 3 Hours](https://hackread.com/fbi-artex-ai-tool-hackers-gain-privileged-access/) — *Hackread – Cybersecurity News, Data Breaches, AI and More*  
Why it matters: The FBI warns that an AI attack framework reaches privileged access in under three hours.  
An FBI flash advisory of 13 November says criminals have used ARTEX, an open-source Go framework that drives agents on DeepSeek, Claude or GPT models to find targets, exploit them and exfiltrate data, since August against finance, government, defence, health, semiconductor and energy targets, with some intrusions reaching privileged access in under three hours; one ConnectWise Cloud Backup compromise reached over 2,250 downstream tenants, and South Korean banks lost data on about 67,000 people. Microsoft links the ConnectWise activity to Storm-0940, which it calls China-aligned.  
severity high · exploited in the wild · EU: NIS2, DORA · actor Storm-0940 (ConnectWise activity) (50%)

**\[P3\]** [The forced audit: how AuditTeam’s OpsAgent framework masks ransomware as consulting](https://securelist.com/auditteam-opsagent-forced-audit/122031/) — *Securelist*  
Why it matters: Ransomware sold as an unrequested 'audit', with victims in Italy and Finland.  
Kaspersky says AuditTeam used its OpsAgent framework, disguised as a pentest tool and rewritten from C++ to Rust, against dozens of organisations in Russia, Italy, the Philippines, Hong Kong, Senegal, Finland and China from January to at least October 2026; it kills security tools with a signed Jiangmin antivirus driver, encrypts files and demands an "Audit & Consulting fee" for a vulnerability report, with C2 on CloudFront subdomains discovered via GitHub.  
severity medium · exploited in the wild · EU: NIS2 · actor AuditTeam (70%)

**\[P3\]** ['TheMailer' Phishing Kit Puts AI in the Driver's Seat](https://www.darkreading.com/perimeter/themailer-phishing-kit-ai-drivers-seat) — *darkreading*  
Why it matters: A phishing kit lets buyers plug in commercial AI models to rewrite every lure past spam filters.  
Varonis says TheMailer, sold for $3,000 to $6,000 in crypto, lets buyers plug in an API key for Claude, ChatGPT, DeepSeek, Grok, Gemini or Mistral to generate and personalise a separate email, subject and attachment for each recipient and rewrite lures per mail provider until they pass spam filters; the seller showed a 25,000-email campaign and a 145,000-target list.  
severity medium · exploited in the wild · EU: NIS2, AI Act

---

## Digital Sovereignty & Identity

[Amazon’s facial recognition safeguards face questions over Israeli military surveillance](https://www.biometricupdate.com/202611/amazons-facial-recognition-safeguards-face-questions-over-israeli-military-surveillance) — *Biometric Update*  
Why it matters: A cloud provider's own limits on facial recognition prove unenforceable once a military is the customer.  
A Guardian investigation found that an Israeli military unit uploaded about 3.5 million photographs of Palestinians to AWS in early 2024 for analysis with Rekognition, after telling Amazon it planned a facial-recognition project, and that soldiers worked from AWS offices in Tel Aviv while exploring uses in Gaza, Biometric Update reports. AWS says it has no visibility into customer workloads and will not discuss the relationship without the customer's consent.

[San Diego dumps both Flock cameras and smart streetlights](https://arstechnica.com/tech-policy/2026/11/san-diego-dumps-both-flock-cameras-and-smart-streetlights/) — *Ars Technica*  
Why it matters: A large US city switches off its licence-plate cameras and smart streetlights.  
San Diego's council voted to end its contracts with Flock and Ubicquia, so more than 500 cameras will go dark three years after the mayor pushed for them; the mayor said he will not veto. A left-right coalition of critics drove the change, and one councilmember switched after saying the cameras made residents feel unsafe. Flock says crimes will take longer to solve.

[UK begins testing DVS for checks at scale, publishes guidance for data sharing](https://www.biometricupdate.com/202611/uk-begins-testing-dvs-for-checks-at-scale-publishes-guidance-for-data-sharing) — *Biometric Update*  
Why it matters: Britain starts testing the plumbing of its digital ID trust framework.  
The UK's Office for Digital Identities and Attributes has begun phased testing of the technology that lets organisations check the register of certified Digital Verification Services securely and at scale, starting with service authentication by mutual TLS or signed client assertions; a second phase adding OpenID Federation is due in two weeks. A draft code of practice for public authorities sharing data with these services has been laid before Parliament.

---

## Defence & National Security

[Pentagon elevates AI as a ‘fundamental requirement’ for cybersecurity staff training](https://defensescoop.com/2026/11/17/pentagon-ai-requirement-cybersecurity-training/) — *DefenseScoop*  
Why it matters: AI becomes a mandatory skill for every Pentagon cyber role.  
Mark Gorak of the Pentagon CIO's office said all cybersecurity roles will now have AI as a fundamental training requirement, as adversaries use models to automate reconnaissance, find zero-days and scale social engineering, and that hiring is shifting to skills tests on cyber ranges rather than credentials.

[NATO aims to spur space industry, help militaries by joining data marketplace](https://www.defenseone.com/threats/2026/11/nato-aims-spur-space-industry-help-militaries-new-data-marketplace/417283/) — *Defense One - All Content*  
Why it matters: NATO buys commercial space-tracking data to protect satellites from Russian manoeuvres.  
NATO will use the Global Data Marketplace run by Colorado-based Bluestaq, already used by the US Space Force, to buy data on space objects from more than 220 companies in 17 allied countries, mixing it with classified data in a separate repository; first orders are expected by year end. Officials cited an October approach by a Russian satellite to one operated by Finland's ICEYE, which supplies radar imagery to Ukraine.

[Space industrial base stretched thin amid low-orbit satellite constellation shift, GAO says](https://defensescoop.com/2026/11/17/space-industrial-base-stretched-thin-gao-report/) — *DefenseScoop*  
Why it matters: The Pentagon's shift to mass small-satellite constellations strains a thin supply chain.  
GAO says the $33 billion five-year shift to constellations of small satellites in low orbit faces shortages of semiconductors, electronics, materials and manufacturing capacity, as 94% of US satellites in orbit are privately owned and the defence base competes with commercial demand; it recommends central oversight of supply-chain gaps across the Air Force and the NRO.

---

## Health & Cyber

[The proposed reform of the Medical Device Regulation and its implications for AI-enabled medical devices in the EU](https://www.nature.com/articles/s41746-026-03475-8) — *npj Digital Medicine*  
Why it matters: Experts say the EU's medical-device reform fixes overlaps but leaves AI devices' problems open.  
A comment in npj Digital Medicine assesses the Commission's December 2025 proposal to amend the MDR, the IVDR and the AI Act in one package. It finds the cuts to double regulation and documentation broadly well-targeted, but says challenges specific to AI-enabled medical devices remain unresolved. Manufacturers have reported delayed launches, discontinued and orphan devices, and some consider leaving the EU market.

**\[P3\]** [🏴‍☠️ Pixelleaks has just published a new victim : FAIR Health, Inc. (fairhealth.com)](https://www.ransomware.live/id/RkFJUiBIZWFsdGgsIEluYy4gKGZhaXJoZWFsdGguY29tKUBQaXhlbExlYWtz) — *Ransomware.live RSS Feed*  
Why it matters: An extortion group claims the internal estate of the US health claims-data nonprofit.  
Pixelleaks lists FAIR Health, the US nonprofit that holds a large national database of healthcare claims, claiming staff user data from its remote-access server (114,449 files), domain and identity material and a map of its cloud environment, with a deadline of 22 November. The claim does not mention patient records and is unverified.  
severity medium · EU: GDPR Art. 9, NIS2 · actor Pixelleaks (40%)

**\[P3\]** [🏴‍☠️ Vortex has just published a new victim : Drive Medical - 1,442,814 records](https://www.ransomware.live/id/RHJpdmUgTWVkaWNhbCAtIDEsNDQyLDgxNCByZWNvcmRzQFZvcnRleA) — *Ransomware.live RSS Feed*  
Why it matters: An extortion group claims 1.4 million records from a medical-equipment maker.  
Vortex lists Drive Medical, a US maker of home medical equipment, claiming 1,442,814 records with a deadline of 20 November; the content of the records and the claim are unverified.  
severity medium · EU: GDPR Art. 9, MDR · actor Vortex (35%)

---

## Quantum & Cryptography

[GSA says post-quantum work on PIV cards shows new encryption can work on old tech](https://cyberscoop.com/gsa-post-quantum-encryption-piv-cards/) — *CyberScoop*  
Why it matters: A US agency puts post-quantum algorithms on the smart cards it already issues.  
GSA ran post-quantum algorithms directly on the chips of its Personal Identity Verification cards, issued new credentials and built the policies, certifications and hardware security modules around them, showing that quantum-resistant credentials work with today's cards, said its deputy associate administrator Dan Pomeroy. Agencies are revising migration plans for the 2030 federal deadline set by executive order; links with weak bandwidth, such as satellites, remain harder.

---

## Cybersecurity & Threats

**\[P1\]** [A vulnpocalypse bug worthy of attention: Researchers disclose critical flaw in Java with help of AI](https://cyberscoop.com/java4shell-critical-java-flaw-openjdk/) — *CyberScoop*  
Why it matters: A flaw in Java itself, found by an AI agent in an hour, reaches Jira, Jenkins and Keycloak.  
Calif Research disclosed Java4Shell (CVE-2026-114938), a memory-corruption flaw in the AVX-512-optimised Base64 decoder of OpenJDK's x86-64 HotSpot JVM: a crafted input writes 64 bytes into a buffer as small as 50, and with heap grooming it gave unauthenticated code execution in Keycloak, Jira, Confluence, Jenkins and PingFederate, and authenticated execution in Kafka, Solr and Elasticsearch; ARM64 is unaffected. Calif says a researcher found it in about an hour with OpenAI's Codex for under $200 and that AI agents built most exploit chains. Oracle and OpenJDK shipped fixes on Tuesday and CISA urged immediate patching; GreyNoise reports scanning and exploitation attempts, with no confirmed compromise yet.  
severity critical (CVSS 9.2) · `CVE-2026-114938` · EU: NIS2, CRA, DORA

**\[P2\]** [Japan takes Fugaku supercomputer offlline after cyberattack](https://www.bleepingcomputer.com/news/security/japan-takes-fugaku-supercomputer-offlline-after-cyberattack/) — *BleepingComputer*  
Why it matters: Japan's flagship supercomputer is shut after intruders stole users' keys and research files.  
RIKEN says an intruder exploited the public status dashboard of the Fugaku supercomputer between 10 and 13 November to reach several internal servers and plant malware aimed at user data, taking names and emails of 4,222 Fugaku and 375 Rikyu users, SSH private keys and access keys, and research data files; some keys were used in attempts to reach other systems. Fugaku was shut down on 14 November and logins disabled; RIKEN has not attributed the attack or found the intrusion method.  
severity high · exploited in the wild · EU: NIS2

**\[P1\]** [SAPBleed: Pre-Auth Memory Disclosure and RCE in SAP NetWeaver, Web Dispatcher and HANA XS (CVE-2026-76964, CVE-2026-89126)](https://seclists.org/fulldisclosure/2026/Nov/36) — *Full Disclosure*  
Why it matters: A pre-authentication flaw in SAP's web layer can leak SSO tickets and TLS keys from inside the network.  
nullFaktor's SAPBleed advisory (CVE-2026-76964 and CVE-2026-89126, CVSS 10.0) describes a heap overflow in SAP's ICM, Web Dispatcher and the Web Dispatcher inside HANA XS: an internet-facing Web Dispatcher can forward a WebSocket attack to an ICM in a protected backend, letting attackers repeatedly read its memory, including SSO tickets and private TLS keys, and reach code execution. SAP Notes 3778530 and 3778556 contain the fixes.  
severity critical (CVSS 10.0) · `CVE-2026-76964` · EU: NIS2, DORA

**\[P2\]** [Attackers Compromise Microsoft's Agent Framework PyPI Package to Steal Cloud Credentials](https://thehackernews.com/2026/11/attackers-compromise-microsofts-agent.html) — *The Hacker News*  
Why it matters: Microsoft's own AI-agent framework on PyPI was turned into a cloud-credential stealer.  
Versions 1.26.1 and 1.26.2 of Microsoft's agent-framework package, published 33 minutes apart on 17 November and since quarantined, dropped a .pth file that runs a stealer on every Python start, even pip; it collected credentials and environment data and asked Azure DevOps, managed-identity endpoints and GitHub Actions to issue new tokens, sending them to a webhook.site address. The package has about 190,000 monthly downloads; how the attacker got publishing access is unknown.  
severity high · exploited in the wild · EU: CRA, NIS2

**\[P3\]** [Apple Services Disrupted as NSA Network Goes Down and Equinor Faces Cyberattack](https://securityaffairs.com/203354/security/apple-outages-add-to-growing-concerns-over-technology-service-disruptions.html) — *Security Affairs*  
Why it matters: Norway's state-controlled energy major confirms a cyberattack on its IT systems.  
Equinor disclosed a cyberattack against some of its IT systems on 17 November, activated its response teams and notified the authorities; no details on impact or attacker were given, Security Affairs reports.  
severity medium · exploited in the wild · EU: NIS2

**\[P3\]** [Cloud keys leaked in Docker Hub images are stolen within seconds](https://www.bleepingcomputer.com/news/security/cloud-keys-leaked-in-docker-hub-images-are-stolen-within-seconds/) — *BleepingComputer*  
Why it matters: Keys left in public container images are used within seconds.  
Stony Brook University researchers published 300 honeypot images on eight registries for four months: the planted AWS keys drew more than 140,000 API calls from 5,000 IP addresses, the first eight seconds after publication and all within 15 minutes, and OpenAI and Anthropic keys were drained. Scanners read the image layers without running the containers; Docker Hub accounted for nearly all the AWS abuse.  
severity medium · exploited in the wild · EU: NIS2, CRA