Weekly review

ThreatNoir Morning Brief — November 17

2026-11-17Morning4 articles
Audio
Listen to the episode

Morning Review in IT Security — November 17, 2026

The cybersecurity landscape faces mounting pressure as supply chain attacks and automated artificial intelligence-driven intrusions reshape threat vectors. On November 16, a critical compromise of the Optimizely content delivery network exposed major global brands to malicious scripts, while security researchers document the emergence of fully autonomous AI cyberattack frameworks deployed by Russian-speaking threat actors.

Optimizely Supply Chain Attack Compromises Major Websites

A significant supply chain attack targeting the Optimizely platform has resulted in malicious JavaScript being injected into the websites of numerous high-profile organizations. The incident affected major brands including Zoom, CNN, Fox News, BBC, Salesforce, and GitLab, exposing their visitors to sophisticated attack payloads. The poisoned Optimizely scripts operated for approximately 17 hours on November 16, during which they performed multiple malicious functions including cryptocurrency address swapping, payment card skimming, and distribution of iOS exploits. Source: Optimizely Malware Hits Zoom, BBC, CNN: What to Do

The attack delivered two primary iOS exploit frameworks identified as Coruna and DarkSword to compromised users. Security researchers have identified malicious infrastructure associated with the campaign, including the domain 1688360.cc and related URLs such as https://1688360.cc/card.php and https://1688360.cc/track.js. The breadth of affected properties and the sophistication of the payload delivery mechanism underscore the critical risk posed by compromised content delivery networks to downstream users. Source: Optimizely supply chain attack infects major brands' sites with malware

Russian-Speaking Actors Deploy Fully Automated AI Cyberattack Framework

Google's Threat Intelligence Group has documented the first fully automated artificial intelligence cyberattack conducted by Russian-speaking threat actors. The attack represents a significant evolution in adversarial capability, as attackers have transitioned from manual prompt-based interactions with chatbots to deploying autonomous multi-agent frameworks capable of executing complex, multi-stage intrusions without human intervention. Source: Google says Russian-speaking hackers ran the first fully automated AI cyberattack

The autonomous AI frameworks identified in active use include ARTEX, CAIRN, CyberStrikeAI, PentAGI, HexStrike AI, and ZeroHollow. These tools prioritize operational speed over stealth, resulting in noisy forensic traces despite their technical success. Google's analysis indicates that threat actors are leveraging autonomous agents to automate exploit deployment, harvest credentials, and scale intrusion capabilities across target environments. The shift toward speed-over-stealth tactics suggests attackers are confident in their ability to establish persistence before defensive teams can respond. Source: Google GTIG Warns AI Agents Automate Exploits, Steal Credentials and Scale Cybercrime

The convergence of supply chain vulnerabilities and automated AI-driven attack frameworks presents organizations with compounded defensive challenges. Immediate patching, supply chain visibility, and behavioral detection capabilities have become essential components of contemporary security operations.

Sources & IOCs

Source articles and their indicators (defanged where appropriate). IP addresses, domains, URLs and file hashes are copied from the publisher's page by pattern matching. Malware names, CVE numbers and technique ids are picked by AI. No person reviews them.

Optimizely Malware Hits Zoom, BBC, CNN: What to Do
Domain4
  • 1688360.cc
    Attacker's server domain used to host malicious script.
  • eil786npto9u5bq.icu
    Written defanged in the article text
  • webshell.lol
    Written defanged in the article text
  • 5vobauua3ojpklx.icu
    Written defanged in the article text
URL1
  • 1688360[.]cc/track[.]js
    Written defanged in the article text
Google GTIG Warns AI Agents Automate Exploits, Steal Credentials and Scale Cybercrime
Malware6
  • CyberStrikeAI
    Autonomous framework connecting commercial and open-weight AI models to 100+ security tools; observed in testing
  • PentAGI
    Autonomous framework designed for less experienced operators; observed in real attacks systematically scanning Asian organizations
  • ARTEX
    Autonomous multi-agent offensive framework used by attackers for multi-stage intrusions
  • ZeroHollow
    Inference service for exploit development and LLM routing used by threat actors
  • HexStrike AI
    LLM-based framework providing access to 150+ tools managed by 12+ specialized agents; observed in testing
  • CAIRN
    AI agentic framework allowing multiple agents to collaborate and share information; observed in real attacks against US aerospace sector