Exfiltration via Cyber Means
Description
Adversaries may exfiltrate AI artifacts or other information relevant to their goals via traditional cyber means.
See the ATT&CK Exfiltration tactic for more information.
- ATLAS ID
- AML.T0025
- Maturity
- RealizedUsed by real adversaries against AI systems in the wild.
- Platforms
- Enterprise
Mitigations
- AML.M0005Control Access to AI Models and Data at Rest
Access controls can prevent exfiltration.
Case studies that use it
- AML.CS0010Microsoft Azure Service Disruption
- AML.CS0015Compromised PyTorch Dependency Chain
- AML.CS0018Arbitrary Code Execution with Google Colab
- AML.CS0023ShadowRay: Hijacking Exposed Ray Clusters
- AML.CS0027Organization Confusion on Hugging Face
- AML.CS0043Malware Prototype with Embedded Prompt Injection
- AML.CS0044LAMEHUG: Malware Leveraging Dynamic AI-Generated Commands
- AML.CS0048Exposed ClawdBot Control Interfaces Leads to Credential Access and Execution
- AML.CS0058Google Photos AI Model Extraction
- AML.CS0059EchoLeak: Zero-Click Prompt Injection Targeting M365 Copilot for Data Exfiltration
- AML.CS0063Prompt-Based Attacks Against Gemini via Calendar Invitations
- AML.CS0068Autonomous OpenAI Evaluation Agents Compromise Hugging Face Infrastructure
- AML.CS0069GTG-1002 Claude Code Espionage Campaign
- AML.CS0071Multi-Agent Framework Compromises Taiwanese Government Systems