Executive Summary
In November 2025, the AI-driven platform JustAskJacky was compromised, exposing sensitive user prompts and data after adversaries exploited weaknesses in encrypted traffic controls and inadequate egress security. Attackers orchestrated lateral movements across internal AI/ML workloads, leveraging insufficient segmentation and lack of effective visibility to siphon proprietary inputs and outputs through encrypted but poorly-monitored channels. The breach remained undetected for weeks, putting affected businesses and consumers at risk of prompt leakage, IP loss, and possible regulatory infractions in industries reliant on AI automation.
This incident underscores a surge in sophisticated AI/ML exploitation techniques and highlights systemic gaps in east-west traffic security, zero trust segmentation, and anomaly detection across multicloud and hybrid environments. As organizations accelerate their adoption of AI-powered platforms, the need for robust compliance and zero trust frameworks has reached critical urgency.
Why This Matters Now
AI-powered environments are increasingly under attack, with adversaries bypassing traditional perimeter defenses by targeting east-west traffic flows and exploiting gaps in AI/ML data handling. The JustAskJacky breach reveals how unchecked shadow AI, poor egress controls, and limited traffic visibility enable large-scale data leakage—making comprehensive zero trust and compliance-aligned security paramount for AI adoption.
Attack Path Analysis
The attacker initiated the compromise by exploiting a misconfigured or vulnerable AI/ML cloud workload, likely gaining access via credential abuse or exposed APIs. They escalated privileges by abusing cloud IAM roles or service identities to acquire broader access. Using east-west movement, the attacker pivoted across workloads or between Kubernetes namespaces to reach sensitive data or additional services. Establishing command and control through covert channels, such as encrypted outbound or misused egress paths, they maintained persistence. Data was exfiltrated from cloud storage or workloads via unsanctioned outbound traffic leveraging encrypted channels. The attack culminated in impactful actions such as ransomware deployment, data encryption, or business disruption, targeting both AI/ML assets and core services.
Kill Chain Progression
Initial Compromise
Description
Compromise of an AI/ML workload through exposed API, misconfiguration, or credential abuse to gain unauthorized cloud access.
Related CVEs
CVE-2024-43572
CVSS 7.8A vulnerability in the Microsoft Management Console (MMC) allows attackers to execute arbitrary code via malicious .msc files.
Affected Products:
Microsoft Windows – 10, 11
Exploit Status:
exploited in the wild
MITRE ATT&CK® Techniques
Phishing
Command and Scripting Interpreter
User Execution
Modify Authentication Process
Valid Accounts
Steal Web Session Cookie
Email Collection
Exfiltration Over C2 Channel
Potential Compliance Exposure
Mapping incident impact across multiple compliance frameworks.
PCI DSS 4.0 – Strong Authentication for Users
Control ID: 8.3.1
NYDFS 23 NYCRR 500 – Cybersecurity Policy
Control ID: 500.03
DORA (EU Digital Operational Resilience Act) – ICT Risk Management Framework
Control ID: Article 9
NIS2 Directive – Cybersecurity Risk Management and Reporting
Control ID: Article 21
CISA Zero Trust Maturity Model (ZTMM) 2.0 – Identity Verification and Authentication
Control ID: Identity Pillar
Sector Implications
Industry-specific impact of the vulnerabilities, including operational, regulatory, and cloud security risks.
Financial Services
AI exploitation threats like JustAskJacky and Rhadamanthys ransomware critically impact financial institutions requiring encrypted traffic protection and zero trust segmentation compliance.
Health Care / Life Sciences
Shadow AI risks and advanced malware threaten patient data integrity, demanding robust egress security, anomaly detection, and HIPAA compliance enforcement.
Information Technology/IT
AI/ML exploitation targeting cloud infrastructure necessitates Kubernetes security, multicloud visibility, and comprehensive threat detection capabilities for service providers.
Government Administration
State-sponsored threats like Salt Typhoon require enhanced east-west traffic security, encrypted communications, and zero trust network architecture implementation.
Sources
- Intelligence Insights: November 2025https://redcanary.com/blog/threat-intelligence/intelligence-insights-november-2025/Verified
- Rhadamanthys Malware: Analysis, Detection, Removal | Huntresshttps://www.huntress.com/threat-library/malware/rhadamanthysVerified
- Police disrupts Rhadamanthys, VenomRAT, and Elysium malware operationshttps://www.bleepingcomputer.com/news/security/police-disrupts-rhadamanthys-venomrat-and-elysium-malware-operations/Verified
- One year later, Rhadamanthys is still dropped via malvertising - ThreatDown by Malwarebyteshttps://www.threatdown.com/blog/one-year-later-rhadamanthys-is-still-dropped-via-malvertising/Verified
Frequently Asked Questions
Cloud Native Security Fabric Mitigations and ControlsCNSF
Comprehensive CNSF and Zero Trust controls—such as segmentation, encrypted traffic enforcement, egress filtering, and continuous threat detection—would have restricted attacker movement, prevented unsanctioned access, and enabled early detection, dramatically containing the impact at multiple kill chain stages.
Control: Zero Trust Segmentation
Mitigation: Attack surface reduced by isolating workloads through identity-based segmentation.
Control: Multicloud Visibility & Control
Mitigation: Misconfigurations and excessive privileges are quickly identified through centralized observability.
Control: East-West Traffic Security
Mitigation: Unauthorized east-west movements are blocked or detected, stalling attacker progression.
Control: Egress Security & Policy Enforcement
Mitigation: Suspicious outbound C2 channels are detected or denied based on FQDN/application policies.
Control: Encrypted Traffic (HPE)
Mitigation: Outbound data is encrypted and observable to prevent unauthorized interception or exfiltration.
Real-time detection of destructive or anomalous activity enables rapid mitigation.
Impact at a Glance
Affected Business Functions
- Finance
- IT Operations
- Customer Support
Estimated downtime: 5 days
Estimated loss: $500,000
Potential exposure of sensitive customer data, including financial records and personal information, due to Rhadamanthys malware infiltration.
Recommended Actions
Key Takeaways & Next Steps
- • Enforce Zero Trust Segmentation to isolate AI/ML and sensitive workloads, blocking unauthorized lateral movement.
- • Implement strict egress policy enforcement and FQDN filtering to disrupt C2 and exfiltration attempts from cloud and Kubernetes environments.
- • Enable high-performance encryption (HPE/MACsec/IPsec) for all data-in-transit and workloads, protecting sensitive data from interception.
- • Leverage continuous threat detection and anomaly response to monitor for covert access, misconfigurations, and runtime abuse across multicloud deployments.
- • Centralize visibility and policy management to rapidly identify privilege escalations and network misconfigurations before attackers can exploit them.
