← Back to ATLAS MatrixTactic 3 of 15

Initial Access

The adversary is trying to get into your AI system.

3 Techniques
MITRE ATLAS Framework

Attack Techniques

Explore the specific techniques adversaries use during the initial access phase of AI attacks.

1
AML.T0010

Exploit Public-Facing Application

Adversaries may exploit public-facing applications to gain initial access to AI systems.

Examples:

  • •Exploiting AI model APIs
  • •Attacking AI web interfaces
  • •Compromising AI service endpoints

Mitigations:

  • ✓Regular security testing
  • ✓Input validation
  • ✓API security controls

References:

MITRE ATLAS AML.T0010
2
AML.T0012

External Remote Services

Adversaries may use external remote services to gain initial access to AI systems.

Examples:

  • •Using VPN services
  • •Leveraging cloud access
  • •Exploiting remote desktop services

Mitigations:

  • ✓Multi-factor authentication
  • ✓Network segmentation
  • ✓Access monitoring

References:

MITRE ATLAS AML.T0012
3
AML.T0015

Hardware Additions

Adversaries may add hardware to gain initial access to AI systems.

Examples:

  • •Adding malicious hardware
  • •Installing rogue devices
  • •Compromising physical infrastructure

Mitigations:

  • ✓Physical security controls
  • ✓Hardware inventory management
  • ✓Regular physical audits

References:

MITRE ATLAS AML.T0015
← Resource DevelopmentATLAS MatrixAI Model Access →