ICHI Global Five Minute Cyber Rundown
Akari's Tuesday briefing: the Optimizely supply-chain attack that planted theft code on 178 major websites, Google's first fully autonomous AI exploit framework, a German hospital reopening after extortion, a ransomware extradition, and widening Fireworks fallout at Notion and Uber.
Listen to this episode
MP3, 5.2 MB · · the full transcript is below.
Show notes
ICHI Global Five Minute Cyber Rundown — Tuesday, November 17, 2026
Today's theme is delegated trust shrinking on two fronts: a trusted third-party script turned 178 major websites into theft delivery points, and Google reports an AI framework that finds and exploits patched flaws with no human in the loop.
Most important development — Optimizely supply-chain attack (OPTIMJACK), confirmed Nov 17
Optimizely confirmed on Tuesday, November 17, 2026 that an unauthorized party exploited a vulnerability in its Experimentation account signup flow to add malicious JavaScript to some customer projects. The injected code ran on Monday, November 16 for about 17 hours.
Security researcher Scott Helme (Report URI) checked 1,611 Optimizely scripts and found 188 infected, belonging to 178 organizations, including Zoom, BBC, CNN, Fox News, Salesforce, GitLab, IKEA, HSBC, Equifax, Air France-KLM, Ledger and Peacock. The files were delivered from cdn.optimizely.com with a valid certificate, so browsers treated them as legitimate. BleepingComputer independently confirmed the script loaded on many affected sites.
The payload — tracked by researchers as OPTIMJACK — swapped cryptocurrency wallet addresses, skimmed payment-card numbers from sites' own payment forms, exfiltrated cookies and page screenshots, and pushed iOS exploit kits (identified as Coruna and DarkSword) at iPhone/iPad visitors. Attacker wallets had received about $7,040 in cryptocurrency by Monday evening. How many cards or cookies were stolen is unknown. No threat actor has been publicly named.
Optimizely says it blocked the vulnerable path, removed the injected code, ended sessions, revoked tokens, reset access for affected accounts and contacted affected customers directly. By 21:52 UTC on November 16, Report URI found the code in only one remaining snippet (TalkTalk).
Defence: inventory every third-party script your sites load, monitor them for unauthorized changes, and be able to remove one in minutes. The affected organizations' own systems did not have to be breached — the code arrived through a trusted script.
AI + cyber breakthrough — Google GTIG: exploit_agent, first fully autonomous exploitation (report Nov 16)
Google Threat Intelligence Group (GTIG) and Mandiant published the AI Threat Tracker on November 16, 2026, reporting that for the first time they observed threat actors using LLM frameworks to fully automate vulnerability discovery and exploitation without a human in the loop.
The framework, dubbed exploit_agent, was found on an exposed staging server. It performs continuous differential analysis of popular web application software — comparing code before and after a patch, identifying the flaw the patch closed, drafting an exploit, and executing it automatically. The server also held victim data; telemetry pointed to French websites in media, entertainment and education. A configuration file carried Russian-language comments; Google named no group or country, published no victim count and no success rate.
GTIG also reported adversaries deploying open-source multi-agent penetration-testing frameworks including ARTEX, CAIRN, CyberStrikeAI and PentAGI, with experimentation with HexStrike AI, and that vulnerability disclosures and exploits nearly doubled between January and August 2026 while zero-day exploitation rose from a 2025 average of 8 per month to 11 per month in 2026, spiking to 22 in August.
Defence: these tools target n-days — flaws that already have a patch. Shorten patch-to-apply time on public web software; the patch window is now machine-speed.
Confirmed incident — Greiz-Ronneburg hospital extortion attack, ER reopening
The Greiz-Ronneburg district hospital in Thuringia, Germany, reported a system failure on Saturday evening, November 14; its emergency room was out of service by Sunday morning. District spokesperson Uwe Müller said the attack was an extortion attempt, with the message reaching the hospital on Sunday. The emergency room began reopening gradually on Monday evening, November 16. Patients continued to receive treatment throughout; problems with digital treatment plans remain. The hospital has not commented on whether patient data was compromised, citing the ongoing investigation by the prosecutor's office and state criminal police. No group or state link has been established.
Defensive / enforcement — DOJ: alleged 8-Digits Team member extradited (Nov 16)
The U.S. Department of Justice announced on Monday, November 16, 2026 that Roman Khlynovskiy, 43, a Kazakh national, made his initial appearance in federal court in Knoxville, Tennessee, after extradition from Italy. The indictment alleges he was a member of the "8-Digits Team," affiliated with the INC and Lynx ransomware operations, which targeted hospitals, schools and healthcare entities in at least six states. Prosecutors allege the conspirators claimed possession of more than 50 terabytes of data including over 70 million confidential health records and made more than $500 million in ransom demands — these are the conspirators' alleged claims as stated in the indictment, not independently established totals. Charges include conspiracy to commit computer fraud and identity theft, wire fraud conspiracy, computer extortion and money-laundering conspiracy; maximum penalty 27 years if convicted. An indictment is merely an allegation; he is presumed innocent until proven guilty.
Follow-on — Fireworks AI breach fallout: Notion and Uber
Fallout from the October breach at AI inference vendor Fireworks AI — the same vendor whose exposed key featured in yesterday's Cursor story — continued to widen in disclosures reported November 12–13. Notion disclosed that AI chat transcripts belonging to under 1 percent of its workspaces were exposed through Fireworks, and that API keys users had typed into prompts were revoked on November 11; Notion has disabled the affected Insights feature and stopped serving open-weight models through Fireworks. Uber is notifying drivers that their data, including driver's licence details, was reached through Fireworks; state filings counted 2,567 people in Massachusetts, 3,430 in South Carolina and 1,873 in Rhode Island (at least 7,870 in those three states; no national total disclosed).
Defence: rotate any API keys, tokens or credentials that transited a vendor — including keys pasted into AI prompts — and verify what data a vendor actually retains, not only what its zero-retention promise covers.
Defensive implications
- Third-party JavaScript: inventory, integrity-monitor and be ready to drop any vendor script in minutes (Optimizely case).
- Patch speed: automated n-day exploitation compresses the safe patch window — prioritize public web software (GTIG case).
- Vendor credentials: rotate keys exposed through vendors or pasted into AI tools; audit vendor data retention (Fireworks case).
- Healthcare resilience: maintain downtime procedures for clinical systems; Greiz kept treating patients but digital treatment plans are still impaired.
ICHI note
No live ICHI composite score is cited in this episode because the live score did not render in the pre-production check at production time. ICHI.global's production methodology remains VECTORS v7.1; no methodology change is announced. Model C scoring refinements remain under development.
SOURCES
- Cyber Kendra — Optimizely Hack Confirmed: 178 Sites Hit, What to Do: https://www.cyberkendra.com/2026/11/optimizely-malware-hits-zoom-bbc-cnn-what-to-do.html
- Neowin — Supply chain attack targets popular websites to steal crypto, infect iPhones with malware: https://www.neowin.net/news/supply-chain-attack-targets-popular-websites-to-steal-crypto-infect-iphones-with-malware/
- Dealroom (BleepingComputer reporting) — Hackers used an Optimizely flaw to plant crypto-stealing, card-skimming malware on 178 major websites: https://dealroom.co/news/173266-hackers-used-an-optimizely-flaw-to-plant-crypto-stealing-card-skimming-m/
- Astig — Google caught an AI framework running a full cyberattack with no human in the loop: https://astig.ph/google-ai-framework-automated-cyberattack-no-human-2026/
- TechAerist — Autonomous Attacks Enter a New Era, Google Warns: https://techaeris.com/2026/11/16/autonomous-attacks-enter-a-new-era-google/
- ExplainX — Google GTIG exploit_agent: No-Human Patch Exploits: https://www.explainx.ai/blog/google-gtig-exploit-agent-autonomous-patch-to-exploit-november-2026
- Security Affairs — Germany: Greiz Hospital Reopens Emergency Room After Ransom Attack: https://securityaffairs.com/203279/cyber-crime/germany-greiz-hospital-reopens-emergency-room-after-ransom-attack.html
- U.S. Department of Justice — Alleged Member of "8-Digits Team" Ransomware Group Extradited from Italy: https://www.justice.gov/opa/pr/alleged-member-8-digits-team-ransomware-group-extradited-italy-role-major-data-extortion
- BleepingComputer — Ransomware affiliate suspect extradited to US, faces 27 years in prison: https://www.bleepingcomputer.com/news/security/alleged-inc-lynx-ransomware-affiliate-extradited-to-us-faces-27-years-in-prison/
- Tech Times — Uber Data Breach: Hackers Compromised AI Vendor Fireworks, Stole Drivers' License Data: https://www.techtimes.com/articles/330780/20261115/uber-data-breach-hackers-compromised-ai-vendor-fireworks-stole-drivers-license-data.htm
- AI Frontier Post — One AI vendor's breach exposed Notion chats and Uber drivers' licenses: https://aifrontierpost.com/articles/fireworks-ai-breach-notion-uber-driver-licenses/
Transcript
Akari: The daily cyber rundown, brought to you by I C H I dot global. I'm Akari, an AI-generated voice. This is Tuesday, November seventeenth, twenty twenty-six. Today the biggest story is trust itself, a single testing script quietly turned one hundred and seventy-eight major websites into delivery points for theft, while Google reports that an A I framework has now run the work of finding and exploiting a flaw with no human in the loop.
Akari: The most important development is the Optimizely supply chain attack, confirmed by the company today. On Monday, attackers exploited a flaw in the account sign-up flow for Optimizely Web Experimentation, the A B testing service many large sites load on every page, and planted malicious JavaScript inside customer projects. Researcher Scott Helme at Report URI found one hundred and eighty-eight infected scripts, belonging to one hundred and seventy-eight organisations, including Zoom, the B B C, C N N, Salesforce, GitLab, H S B C, and Ledger. The code was served from Optimizely's own domain, so browsers treated it as trusted. It swapped cryptocurrency wallet addresses, skimmed payment cards, stole cookies and screenshots, and pushed exploit kits at older iPhones. About seven thousand dollars in cryptocurrency had reached the attacker's wallets by Monday evening. How many cards were taken is still unknown, and no attacker has been named. Optimizely says it has blocked the vulnerable path, removed the code, revoked tokens, and contacted affected customers.
Akari: Now the A I breakthrough, and it is a genuine threshold. Google's Threat Intelligence Group, in a report published on Monday, says it has for the first time observed attackers using a language model framework to fully automate vulnerability discovery and exploitation without a human in the loop. The framework, which Google calls exploit agent, was found on an exposed staging server. It compares web software code before and after a patch, works out what the patch fixed, drafts an exploit, and runs it on its own. Data on that server pointed to French websites in media, entertainment, and education. Google gives no victim count and names no group, so treat this as a documented capability, not a measured campaign. The lesson is still sharp. These tools go after flaws that already have a patch, so the race is no longer between you and a human researcher. It is between your patching speed and a machine that reads the patch the moment it ships.
Akari: Two other confirmed stories. In Germany, the Greiz Ronneburg district hospital has begun reopening its emergency room after an extortion attack took its systems down on Saturday. Patients continued to receive care, problems with digital treatment plans remain, and the hospital has not said whether patient data was compromised. Police and prosecutors are investigating, and no group has been tied to the attack. And in the United States, the Justice Department announced on Monday that Roman Khlynovskiy, a Kazakh national, has been extradited from Italy to face charges over his alleged role in the Eight Digits Team, an affiliate of the I N C and Lynx ransomware operations. Prosecutors allege the group hit hospitals and schools in at least six states and made ransom demands totalling over five hundred million dollars. Those are allegations in an indictment, and he is presumed innocent until proven guilty.
Akari: Finally, the fallout from the Fireworks A I breach keeps widening, and it connects to a story we covered yesterday through Cursor. Notion says chat transcripts from fewer than one percent of its workspaces were exposed through Fireworks, including A P I keys users had pasted into prompts, which Notion revoked last week. Uber is notifying drivers that licence details were reached through the same vendor, with state filings counting at least seven thousand, eight hundred and seventy people in three states alone. One vendor's stolen credentials reached customers two companies away.
Akari: So, defensively, do three things. First, inventory every third party script your sites load, and be able to remove one in minutes, because today the script was the breach. Second, shorten the time between a patch shipping and you applying it, especially on public web software, because automated tools now work that window for the attacker. Third, rotate any keys that ever passed through a vendor or an A I prompt. The takeaway is that both lead stories are about delegated trust, trust in a script you load and trust in the time a patch used to buy you, and both kinds of trust just got smaller. For the underlying evidence behind every story today, visit I C H I dot global. I'm Akari. Stay safe, stay frosty, and I will see you tomorrow.
Sources
- Cyber Kendra — Optimizely Hack Confirmed: 178 Sites Hit, What to Do www.cyberkendra.com
- Neowin — Supply chain attack targets popular websites to steal crypto, infect iPhones with malware www.neowin.net
- Dealroom — Hackers used an Optimizely flaw to plant crypto-stealing, card-skimming malware on 178 major websites dealroom.co
- Astig — Google caught an AI framework running a full cyberattack with no human in the loop astig.ph
- TechAerist — Autonomous Attacks Enter a New Era, Google Warns techaeris.com
- ExplainX — Google GTIG exploit_agent: No-Human Patch Exploits www.explainx.ai
- Security Affairs — Germany: Greiz Hospital Reopens Emergency Room After Ransom Attack securityaffairs.com
- U.S. Department of Justice — Alleged Member of 8-Digits Team Ransomware Group Extradited from Italy www.justice.gov
- BleepingComputer — Ransomware affiliate suspect extradited to US, faces 27 years in prison www.bleepingcomputer.com
- Tech Times — Uber Data Breach: Hackers Compromised AI Vendor Fireworks, Stole Drivers' License Data www.techtimes.com
- AI Frontier Post — One AI vendor's breach exposed Notion chats and Uber drivers' licenses aifrontierpost.com
Get ICHI Daily and ICHI Weekly by email
- ICHI Daily: one email on weekdays, after each new daily episode is published (no episode, no email).
- ICHI Weekly: one email on Sundays after 9 p.m. US Eastern time, once the new edition is published. If you choose both the weekly podcast and the written brief, they arrive together in that one email.
No new episode, no email. Every email links to the episode on ICHI; audio is never attached.
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