Executive Summary
Between August and October 2026, the Russia-linked threat actor APT29 conducted a sophisticated multi-vector mobile exploitation campaign dubbed 'Poisoned Source.' Attackers compromised legitimate websites, registered lookalike domains, and injected malicious JavaScript into the Plerdy SEO Tracker supply chain (affecting the widely loaded script at a.plerdy[.]com) to deliver platform-specific N-day exploit chains to iOS and Android users. iOS victims were targeted with the DarkSword exploit kit across CVE-2025-43529, CVE-2026-20700, and related vulnerabilities, enabling full device data exfiltration including iMessages, iCloud Keychain, and Wi-Fi passwords. Android users were served a Chrome-targeting chain exploiting CVE-2026-3910 and CVE-2025-27038, culminating in deployment of the AndroSnitch infostealer.
This campaign illustrates the accelerating commoditization of commercial surveillance vendor exploit frameworks. The confirmed reuse of DarkSword across both nation-state and criminal actors, combined with the Coruna kit's broad availability following public disclosure, signals a structural shift: advanced mobile exploitation is no longer exclusive to well-resourced state programs, raising the risk floor for enterprises, government agencies, and critical infrastructure operators globally.
Why This Matters Now
Mobile endpoints have become primary targets for nation-state and criminal actors alike, yet most enterprise security architectures lack visibility into mobile traffic flows and browser-level exploit delivery. With DarkSword and Coruna kits now accessible to multiple threat actors and distributed via supply chain compromise, expired analytics domains, and malvertising, organizations face a rapidly widening attack surface that traditional perimeter and endpoint controls were not designed to address.
Attack Path Analysis
APT29-linked actors compromised the Plerdy SEO Tracker supply chain and injected malicious JavaScript into legitimate websites and lookalike domains, profiling visiting mobile devices and delivering platform-specific N-day exploit chains (DarkSword for iOS, CVE-2026-3910/CVE-2025-27038 for Android). Once exploitation succeeded, payloads harvested cookies, credentials, SMS, iCloud data, and Wi-Fi passwords, exfiltrating all collected data to a hardcoded attacker-controlled telemetry endpoint, ultimately compromising high-value targets including military, government, and defense-adjacent visitors.
Kill Chain Progression
This analysis maps confirmed threat intelligence to the full cloud kill chain to show where defensive gaps would emerge as an attack progresses.
Initial Compromise
Description
Attackers injected malicious JavaScript into the Plerdy SEO Tracker third-party analytics script (a.plerdy[.]com/public/js/click/main.js) between August 27 and September 3, 2026, and also directly compromised legitimate websites across the U.S., Ukraine, Czech Republic, and Bulgaria, as well as registering lookalike domains. Visiting mobile users unknowingly loaded the malicious script, which fingerprinted their browser and device to determine exploit eligibility. Additionally, expired analytics and advertising domains were re-registered under the DeadTag scheme to serve malicious JavaScript to iPhone visitors at scale.
Related CVEs
CVE-2024-27834
CVSS 5.5A vulnerability in Apple WebKit allows an attacker with arbitrary read and write capability to bypass Pointer Authentication, potentially enabling arbitrary code execution on affected iOS devices.
Affected Products:
Apple iOS – < 16.7.8, 16.7.x
Apple Safari – < 17.5
Exploit Status:
exploited in the wildCVE-2024-5274
CVSS 9.6A type confusion vulnerability in Google Chrome's V8 JavaScript engine that allows a remote attacker to exploit heap corruption via a crafted HTML page, leading to arbitrary code execution.
Affected Products:
Google Chrome – < 125.0.6422.112
Exploit Status:
exploited in the wildCVE-2024-4671
CVSS 9.6A use-after-free vulnerability in the Visuals component of Google Chrome that allows a remote attacker to exploit heap corruption via a crafted HTML page, leading to potential arbitrary code execution.
Affected Products:
Google Chrome – < 124.0.6367.201
Exploit Status:
exploited in the wildCVE-2025-27038
CVSS 7.5A memory safety vulnerability in Qualcomm GPU driver components that can be exploited via a malicious application or browser-based attack to achieve privilege escalation on affected Android devices.
Affected Products:
Qualcomm Snapdragon GPU Driver – Multiple Snapdragon chipsets prior to June 2025 security patch
Exploit Status:
exploited in the wild
MITRE ATT&CK® Techniques
Drive-by Compromise
Compromise Software Supply Chain
Command and Scripting Interpreter: JavaScript
Steal Web Session Cookie
Data from Local System
Exfiltration Over C2 Channel
Exploitation for Client Execution
Obfuscated Files or Information
Potential Compliance Exposure
Mapping incident impact across multiple compliance frameworks.
NIST SP 800-53 Rev 5 – Malicious Code Protection / Spam and Phishing Protection
Control ID: SI-3 / SI-8
CISA Zero Trust Maturity Model 2.0 – Application Integrity Verification and Device Posture Enforcement
Control ID: Application Pillar — Visibility & Analytics / Device Pillar — Device Compliance
NIS2 Directive (EU) 2022/2555 – Supply Chain Security and Vulnerability Management
Control ID: Article 21 — Cybersecurity Risk Management Measures
PCI DSS 4.0 – Payment Page Script Integrity and Change Detection Monitoring
Control ID: Requirement 6.4.3 / Requirement 11.6.1
GDPR (EU) 2016/679 – Technical Safeguards and Personal Data Breach Notification
Control ID: Article 32 — Security of Processing / Article 33 — Notification of Breaches
DORA (EU) 2022/2554 – Third-Party ICT Service Provider Risk Assessment and Monitoring
Control ID: Article 28 — ICT Third-Party Risk Management
Sector Implications
Industry-specific impact of the vulnerabilities, including operational, regulatory, and cloud security risks.
Government Administration
APT29's Poisoned Source supply-chain and watering hole campaign directly targets government-adjacent websites, risking exfiltration of cookies, iMessages, and iCloud credentials via DarkSword and AndroSnitch mobile exploit chains.
Military Industry
Sekoia confirms army, special forces, and UAV/ballistic equipment sites were specifically compromised, exposing military personnel mobile devices to full data exfiltration through CVE-2026-20700 and CVE-2025-43529 exploit chains.
Chemicals
A major U.S. chemical industry trade association website was explicitly compromised as a watering hole, placing member organization mobile users at high risk of credential and session cookie theft via N-day iOS and Android exploits.
Information Technology/IT
Supply-chain compromise of Plerdy SEO Tracker JavaScript injected malicious code across thousands of sites, exposing IT professionals to lateral movement risk and egress exfiltration gaps mapped to NIST.800-53.SC-7 and PCI.4.0.3.5.1 compliance failures.
Sources
- APT29-Linked Hackers Hijack Web Analytics Script to Deliver iOS and Android Exploitshttps://thehackernews.com/2026/11/apt29-linked-hackers-hijack-web.htmlVerified
- Poisoned Source: APT29 Watering Holes Target Mobile Devices with N-Day Exploitshttps://www.sekoia.io/en/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits/Verified
- ANDROSNATCH and Watering Hole Campaigns Targeting Android Users - Google TAGhttps://blog.google/threat-analysis-group/tag-bulletin-q3-2024/Verified
- CVE-2024-27834 NVD Detailhttps://nvd.nist.gov/vuln/detail/CVE-2024-27834Verified
- CVE-2024-5274 NVD Detail - Chrome V8 Type Confusionhttps://nvd.nist.gov/vuln/detail/CVE-2024-5274Verified
- CVE-2024-4671 NVD Detail - Chrome Visuals Use-After-Freehttps://nvd.nist.gov/vuln/detail/CVE-2024-4671Verified
- CISA Known Exploited Vulnerabilities Cataloghttps://www.cisa.gov/known-exploited-vulnerabilities-catalogVerified
- Inside Coruna and DarkSword Chains: From Web Lures to iOS Exploitation - Zimperium zLabshttps://zimperium.com/blog/inside-coruna-and-darksword-chains-from-web-lures-to-ios-exploitationVerified
- DeadTag: How Forgotten Website Tags Became a Delivery Network for iPhone Spyware - Report URIhttps://blog.report-uri.com/deadtag-how-forgotten-website-tags-became-a-delivery-network-for-iphone-spyware/Verified
- Malvertising: Coruna and DarkSword iPhone Exploits - iVerifyhttps://www.iverify.io/blog/malvertising-coruna-darksword-iphone-exploitsVerified
- APT29 Profile - MITRE ATT&CKhttps://attack.mitre.org/groups/G0016/Verified
Frequently Asked Questions
Cloud Native Security Fabric Mitigations and ControlsCNSF
Based on the attack progression modeled above, these are the defensive controls that would constrain each stage.
Aviatrix Zero Trust CNSF is relevant here because this supply chain watering hole campaign relied on unrestricted outbound paths and uncontrolled east-west reachability to propagate exploit payloads and exfiltrate harvested credentials from compromised devices. By enforcing segmented egress policies and workload-level access controls across cloud-hosted environments, CNSF would likely reduce the blast radius available to attackers operating through compromised third-party analytics infrastructure.
Control: Cloud Native Security Fabric (CNSF)
Mitigation: CNSF fabric-level visibility across cloud-hosted web properties would likely constrain the reachability of injected third-party scripts to sensitive internal workloads, reducing the scope of what compromised analytics infrastructure could touch within protected cloud environments.
Control: Zero Trust Segmentation
Mitigation: Zero Trust segmentation enforced at the workload and identity layer would likely reduce the privilege scope available to exploit chains attempting to pivot from browser-level execution into protected data tiers or cloud-connected backend services.
Control: East-West Traffic Security
Mitigation: East-west traffic enforcement would likely constrain the reachability of sessions originating from compromised personal devices into cloud-hosted enterprise workloads, reducing how far harvested credentials could be leveraged to move across segmented cloud environments.
Control: Multicloud Visibility & Control
Mitigation: Multicloud visibility and control applied across egress paths would likely reduce the ability of malicious JavaScript to reach hardcoded attacker-controlled telemetry endpoints undetected, constraining the command and control reachability available to payload delivery infrastructure.
Control: Egress Security & Policy Enforcement
Mitigation: Egress security policy enforcement would likely reduce the available outbound paths for bulk data exfiltration to hardcoded external telemetry endpoints, constraining the volume and reach of credential and sensitive data transfers leaving cloud-adjacent environments.
Even with prior-stage controls applied, residual risk would remain around harvested device credentials that exist entirely outside cloud enforcement boundaries, though the blast radius of follow-on enterprise intrusions would likely be meaningfully reduced by segmented access policies constraining what those credentials could reach.
Impact at a Glance
Affected Business Functions
- Mobile Device Security and Endpoint Integrity
- Sensitive Government and Military Communications
- Diplomatic and Intelligence Personnel Data Protection
- Supply Chain Software Integrity (Web Analytics Services)
Estimated downtime: N/A
Estimated loss: N/A
Highly sensitive data exfiltrated from compromised iOS and Android mobile devices of high-value targets including military personnel, government officials, diplomatic staff, and individuals associated with defense and chemical industry organizations across the U.S., Ukraine, Czech Republic, and Bulgaria. Exposed data categories include: SMS messages and iMessages, Apple Mail databases, voicemail, WhatsApp and Telegram message databases, iCloud Keychain credentials, Wi-Fi passwords, contact lists, call history, Safari browser cookies and history, photos, Apple Health records, Calendar entries, iCloud Drive files, and Chrome browser cookies and session data on Android. The Plerdy SEO Tracker supply chain compromise (August 27 to September 3, 2026) introduced the malicious JavaScript loader to all websites using the analytics service, potentially extending exposure beyond the directly targeted high-value sites to any visitor using a vulnerable iOS or Android device during that window.
Recommended Actions
Key Takeaways & Next Steps
- • Deploy Egress Security & Policy Enforcement with FQDN filtering to block outbound connections to unrecognized or newly registered domains such as ad.starx-net[.]com and ad-track[.]live; enforce allowlist-based egress policies so that workloads and managed endpoints can only reach approved telemetry and analytics destinations, preventing exfiltration to hardcoded attacker endpoints.
- • Enable Inline IPS (Suricata) with up-to-date signatures covering known N-day exploit traffic patterns associated with DarkSword, Coruna, CVE-2026-3910, and CVE-2025-27038; inline inspection of outbound JavaScript delivery and exploit payload traffic can identify and block malicious payloads before device compromise occurs where traffic is not end-to-end encrypted at the mobile layer.
- • Apply Zero Trust Segmentation and Cloud Firewall (ACF) controls to restrict third-party analytics and supply chain JavaScript origins; enforce content-security-policy-equivalent network-layer controls that limit which external script domains can be loaded by internal web properties, reducing the attack surface of supply chain injection via compromised CDN or analytics providers like Plerdy.
- • Leverage Multicloud Visibility & Control with anomaly detection to baseline expected egress traffic patterns from web-facing assets and alert on novel high-volume POST requests to non-standard ports (e.g., port 8881) or newly observed external domains; centralized traffic observability across multi-cloud and hybrid environments enables rapid detection of C2 telemetry beaconing indicative of watering hole payload delivery.
- • Enforce Encrypted Traffic (HPE) inspection policies and East-West Traffic Security controls to ensure that data in transit between compromised endpoints and internal resources cannot be silently intercepted; combine with device posture validation at network access points so that devices exhibiting indicators of mobile exploitation (unexpected credential store access, anomalous outbound data volumes) are quarantined before harvested credentials can be used for lateral movement into enterprise or government network segments.



