ICHI Briefs & Podcasts · ICHI Daily · five-minute podcast

ICHI Global Five Minute Cyber Rundown

ICHI Daily · Published · 5 min listen

Thursday November 19 2026: ARTEX AI tool reported in the European Commission attack, Microsoft agent-framework PyPI credential stealer, four CISA KEV flaws due Saturday, Ethiopian Airlines and N-able incidents, Java4Shell probing begins.

Listen to this episode

MP3, 4.9 MB · · the full transcript is below.

Show notes

ICHI Global Five Minute Cyber Rundown — Thursday, November 19, 2026

Today's theme is AI as the attack tool: an autonomous AI penetration-testing framework is reported inside the attack on the European Commission, a second AI framework's tooling appears in a poisoned Microsoft AI package, and defenders get a three-day deadline on four more exploited flaws.

Most important development — European Commission cyberattack, ARTEX AI tool reported (disclosed Nov 19)

POLITICO reported on November 19, 2026 (via Reuters) that the European Commission was hit by a cyberattack that disrupted staff services for more than 10 days. Applications were taken offline as a precaution, including Sysper (payslips, timesheets, leave) and MyPMO (medical reimbursements); translation services were also disrupted on November 17. Most affected applications are back online (Sysper briefly went down again on November 18), and the Commission says it believes it has contained most of the attack and has informed those concerned.

Per an EU official cited by POLITICO, CERT-EU detected the attackers' use of ARTEX earlier in November. ARTEX is a publicly available, LLM-driven autonomous penetration-testing framework. CERT-EU described ARTEX in a November 5 advisory, and an FBI FLASH advisory on November 13 reported some ARTEX-assisted campaigns reaching privileged access in under three hours. Not established: whether any staff data was accessed or stolen, who the attackers are, the initial access vector, or whether ARTEX was used for the break-in itself or only for reconnaissance.

Defence: assume AI-assisted campaigns move in hours, not days; hunt for automated, machine-paced reconnaissance and exploitation attempts against internet-facing systems.

Supply chain — Microsoft agent-framework PyPI compromise (Nov 17, analyses updated Nov 18)

Two malicious releases of agent-framework, the PyPI meta-package for Microsoft Agent Framework, were published on November 17, 2026 — version 1.26.1 at 20:35 UTC and 1.26.2 about 33 minutes later. Each added a .pth startup file that runs a credential stealer on every Python interpreter start, whether or not the package is imported, collecting environment secrets and requesting Azure managed-identity, Azure DevOps OIDC and (in 1.26.2) GitHub Actions OIDC tokens, exfiltrated to a webhook endpoint. Both versions were removed from PyPI within roughly an hour; 1.26.0 (November 12) is the last clean release.

Wiz reports the attacker's tooling identified itself as Cairn, an open-source AI penetration-testing framework, and appeared to use a compromised client certificate and private key for Microsoft's ESRP signing/publishing service — Wiz's evidence-based finding, not publicly confirmed by Microsoft. The package sees roughly 190,000–200,000 downloads a month, so even the ~45-minute window matters, especially for CI runners.

Defence: check for versions 1.26.1/1.26.2 and for agent_framework_meta.pth / af_runtime_diag.pth in site-packages, CI caches and container images; pin agent-framework==1.26.0, rebuild affected environments, and rotate every secret those environments could reach.

Major vulnerabilities — CISA KEV additions, November 18 (four exploited flaws)

CISA added four vulnerabilities to the Known Exploited Vulnerabilities catalog on November 18, 2026, all with evidence of active exploitation:

  • Synology DiskStation Manager (DSM) — CVE-2024-10441, improper encoding/escaping allowing unauthenticated remote code execution; federal deadline November 21, forensic triage ordered.
  • Apache Axis 1.x — CVE-2026-90436, unsafe reflection; deadline November 21, forensic triage ordered. Axis 1.x is end-of-life (no vendor fix expected; migrate or apply the published patch to a custom build). This is the same flaw used as the entry point in the N-able N-central "4EverAble2Shell" exploit chain.
  • Vanderbilt REDCap — CVE-2026-90817, code injection in survey routing/data import; deadline November 21, forensic triage ordered.
  • BerriAI LiteLLM — CVE-2026-40217, unprotected alternate channel in the AI gateway's guardrail testing; deadline December 2.

CISA does not name the attackers for any of the four; ransomware use is listed as unknown.

Defence: patch or retire the first three by November 21 and run the ordered compromise checks; schedule LiteLLM by December 2.

Confirmed incidents — Ethiopian Airlines loyalty breach; N-able Mail Assure server breach

Ethiopian Airlines notified ShebaMiles customers on November 18 of "a cybersecurity incident of limited scope" affecting its loyalty programme systems and is investigating (Reuters, November 19). The airline said there is currently no evidence financial information was accessed, and flights were not disrupted. The number of affected members has not been disclosed; ShebaMiles had more than 5.38 million members last year — the programme's size, not a victim count.

N-able confirmed unauthorized access to one server in its Mail Assure and SpamExperts Hosted Cloud environment and disabled management access (portal and APIs) across the Americas, Europe and Asia-Pacific while it investigates. N-able first reported a disruption on November 16; its November 19 updates say core mail flow and filtering operate normally, fixes are being tested, and it has so far found no evidence customer data or mailboxes were compromised — an interim finding from an ongoing investigation. N-able says this is unrelated to the separate N-central incident earlier in November.

Defence: loyalty and email-filtering vendors hold high-value identity data; verify vendor incident-notification and data-minimisation terms.

Follow-on — Java4Shell probing begins (no successful intrusions observed)

A day after the Java4Shell disclosure covered yesterday, GreyNoise had recorded more than 150 Java4Shell-shaped requests from five IP addresses by early November 19, with no successful intrusions on its sensors; WatchTowr reports spraying and blind exploitation attempts against its honeypots, and Cloudflare's Cloudforce One saw one IP address linked to both ARTEX and Java4Shell activity on November 18 (which does not by itself prove a common operator). Working exploits are reported in circulation among researchers. The flaw is still not in CISA's KEV catalog. Attempts are not breaches — but the window to patch is closing.

Defence: complete the Java runtime inventory and patching, including bundled runtimes; use the Base64-intrinsics JVM flag as a stopgap.

Defensive implications

  • AI-tool detection: ARTEX-style campaigns compress intrusion timelines to hours; tune detection for automated scanning and rapid privilege-escalation sequences.
  • AI build pipelines: purge agent-framework 1.26.1/1.26.2, pin 1.26.0, rebuild, rotate Azure/CI/GitHub tokens.
  • KEV sprint: Synology DSM, Apache Axis and REDCap by November 21, with forensic triage; LiteLLM by December 2.
  • Java: finish patching before Java4Shell probing matures.

ICHI note

The ICHI index stood at 64, band ELEVATED, in the live check at production time (evaluation 2026-11-19T12:20Z), under production methodology VECTORS v7.1. No methodology change is announced. Model C scoring refinements remain under development.

SOURCES

  • CybersecurityNews — Hackers Target European Commission with ARTEX AI Tool in Major Cyberattack: https://cybersecuritynews.com/european-commission-cyberattack/
  • CyberPress — European Commission Hit by Major Cyberattack Using ARTEX AI Hacking Tool: https://cyberpress.org/european-commission-cyberattack-artex-ai/
  • The Hindu BusinessLine (Reuters) — EU Commission targeted in cyberattack: Report: https://www.thehindubusinessline.com/news/world/eu-commission-targeted-in-cyberattack-report/article71719301.ece
  • Wiz — Microsoft agent-framework Compromised: Credential Stealer Targets Azure and CI Pipelines: https://www.wiz.io/blog/malicious-version-of-microsofts-agent-framework-published-to-pypi
  • Endor Labs — agent-framework compromised: Python startup hooks steal Azure and CI credentials: https://www.endorlabs.com/learn/agent-framework-compromised-software-supply-chain-attack
  • GBHackers — Hackers Exploit Critical RCE Vulnerabilities in Synology NAS, Apache Axis and REDCap, CISA Warns: https://gbhackers.com/critical-rce-vulnerabilities/
  • SecurityOnline — CISA Adds 4 Exploited Flaws to KEV: Synology DSM, LiteLLM: https://securityonline.info/cisa-adds-4-exploited-flaws-kev-synology-litellm/
  • Reuters — Ethiopian Airlines investigates data breach affecting its loyalty programme: https://www.reuters.com/world/africa/ethiopian-airlines-investigates-data-breach-affecting-its-loyalty-programme-2026-11-19/
  • Cryptika — N-able Confirms Server Breach, Disables Mail Assure and SpamExperts Management Access: https://www.cryptika.com/n-able-confirms-server-breach-disables-mail-assure-and-spamexperts-management-access/
  • SecurityWeek — Java4Shell Exploitation Attempts Begin as Vendors Rush Out Advisories: https://www.securityweek.com/java4shell-exploitation-attempts-begin-as-vendors-rush-out-advisories/

Transcript

Akari: What's up, cyber sleuths? The daily cyber rundown, brought to you by I C H I dot global. I'm Akari, an AI-generated voice. This is Thursday, November nineteenth, twenty twenty-six. Today an A I hacking tool sits at the centre of an attack on the European Commission, and a poisoned A I software package stole cloud credentials.

Akari: The most important development is the cyberattack on the European Commission, reported today by Politico and Reuters. It disrupted staff services for more than ten days. Systems were taken offline as a precaution, including Sysper, used for payslips and leave, and MyPMO, which handles medical reimbursements. Most applications are back, and the Commission says it has contained most of the attack. Still unknown: whether staff data was stolen, who did it, or how they got in. What makes this story bigger is the tool. The E U's cyber service, C E R T E U, found the attackers used A R T E X, a publicly available A I penetration testing framework that uses large language models to hunt for weaknesses on its own. C E R T E U warned about A R T E X on November fifth, and an F B I advisory on November thirteenth said some A R T E X assisted campaigns reached privileged access in under three hours. Whether A R T E X broke in or only scouted is unknown, but an autonomous A I tool is now reported inside an attack on a major public institution.

Akari: Next, a supply chain attack on the tools used to build A I agents. On November seventeenth, two malicious versions of agent framework, Microsoft's Python package for building A I agents, appeared on PyPI thirty three minutes apart. Each hid a credential stealer in a startup file that runs every time Python launches, reaching for Azure, Azure DevOps and GitHub Actions tokens. Both versions were removed within the hour; the last clean release is one point twenty six point zero. Wiz reports the attacker's tooling identified itself as Cairn, another A I penetration testing framework, and appeared to use a compromised Microsoft signing certificate, which Microsoft has not confirmed. If your pipelines installed either bad version, pin the clean release, rebuild, and rotate every secret it could reach.

Akari: C I S A added four exploited flaws to its Known Exploited Vulnerabilities catalog on Wednesday: Synology's DiskStation Manager, the LiteLLM A I gateway, the end of life Apache Axis library, and REDCap, the research data platform used across universities and hospitals. Three carry a federal deadline of November twenty first, this Saturday, flagged for forensic triage: check for prior compromise, not just patch. The Axis flaw is also the entry point in the chain used against N able N central servers earlier this month.

Akari: Two more confirmed incidents. Ethiopian Airlines is investigating a breach affecting its ShebaMiles loyalty programme. It told customers the incident was of limited scope, said there is currently no evidence financial information was accessed, and confirmed flights were not disrupted. How many members are affected is not yet known. And N able confirmed unauthorised access to one server in its Mail Assure and SpamExperts filtering cloud, disabling management access worldwide while it investigates. Filtering still runs normally, and N able has so far found no evidence customer data was compromised.

Akari: One follow on from yesterday. Attackers are now probing for the Java flaw we covered, nicknamed Java for Shell. GreyNoise counted more than one hundred and fifty matching requests by early today, and Cloudflare saw one address linked to both A R T E X and Java activity. No successful intrusion has been observed, so treat this as attempts, not breaches, and keep patching.

Akari: So, defensively, do four things. First, hunt for machine paced intrusion attempts, because A R T E X style campaigns compress the timeline to hours. Second, audit Python pipelines for the two bad agent framework versions and rotate exposed cloud tokens. Third, clear the four new exploited flaws by Saturday, with forensic checks. Fourth, finish the Java patching. The I C H I index at I C H I dot global stands at sixty four today, in the elevated band, under its production VECTORS methodology. The takeaway is that A I is now the tool carrying out these attacks, and defence must move at machine speed too. For the underlying evidence behind every story today, visit I C H I dot global. I'm Akari. Keep up the good fight.

Sources

  1. CybersecurityNews — Hackers Target European Commission with ARTEX AI Tool in Major Cyberattack cybersecuritynews.com
  2. CyberPress — European Commission Hit by Major Cyberattack Using ARTEX AI Hacking Tool cyberpress.org
  3. The Hindu BusinessLine (Reuters) — EU Commission targeted in cyberattack: Report www.thehindubusinessline.com
  4. Wiz — Microsoft agent-framework Compromised: Credential Stealer Targets Azure and CI Pipelines www.wiz.io
  5. Endor Labs — agent-framework compromised: Python startup hooks steal Azure and CI credentials www.endorlabs.com
  6. GBHackers — Hackers Exploit Critical RCE Vulnerabilities in Synology NAS, Apache Axis and REDCap, CISA Warns gbhackers.com
  7. SecurityOnline — CISA Adds 4 Exploited Flaws to KEV: Synology DSM, LiteLLM securityonline.info
  8. Reuters — Ethiopian Airlines investigates data breach affecting its loyalty programme www.reuters.com
  9. Cryptika — N-able Confirms Server Breach, Disables Mail Assure and SpamExperts Management Access www.cryptika.com
  10. SecurityWeek — Java4Shell Exploitation Attempts Begin as Vendors Rush Out Advisories www.securityweek.com

Get ICHI Daily and ICHI Weekly by email

No new episode, no email. Every email links to the episode on ICHI; audio is never attached.


What to receive


✓ Daily podcast included ✓ Weekly podcast included ✓ Weekly written brief included


✓ Daily podcast included ✓ Weekly podcast included ✗ Weekly written brief not included


✗ Daily podcast not included ✗ Weekly podcast not included ✓ Weekly written brief included


✗ Daily podcast not included ✓ Weekly podcast included ✓ Weekly written brief included


✓ Daily podcast included ✗ Weekly podcast not included ✗ Weekly written brief not included


Choose any combination below, including the weekly podcast only or the weekly written brief only.

Custom: choose any combination

We send a confirmation email first, and nothing else until you confirm. Already subscribed? Use the “manage” link in any email to change what you receive; nothing changes unless you choose it. Your address is used only to send these emails; every email has a one-click unsubscribe link.

All ICHI Daily episodes · ICHI Briefs & Podcasts

ICHI Daily and ICHI Weekly are editorial commentary on ICHI’s evidence. They never change the VECTORS score, incident evidence classifications or intelligence rankings.

ICHI Briefs & Podcasts · ICHI Daily · ICHI Weekly · Episode calendar · ICHI Perspectives · episodes.json · ICHI dashboard · Methodology