Automated Data Analysis

When AI systems process customer data en masse, a highly attractive target for data theft and espionage is created.

Your Strategy

Anonymization is the key here. Sensitive data should be masked or protected using differential privacy before it is passed to the AI model.

Best Practices

▸Differential Privacy: Add controlled noise to the data to prevent identification of individuals.
▸Audit-Proof: Log every access to the training or analysis dataset in an unalterable manner.
▸Minimization: Process only the data fields that are absolutely necessary for the specific analysis.

STRIDE-LM Design Risks

SpoofingS-01

Identity Spoofing

Attacker impersonates a legitimate user or partner.

TamperingT-AI-02

Training Data Poisoning

Manipulation of training data to influence model behavior.

RepudiationR-01

Audit Log Manipulation

Deleting or altering traces of an action.

Information DisclosureI-AI-02

Model Inversion / Membership Inference

Reconstructing training data from model responses.

Denial of ServiceD-01

Resource Exhaustion

Overloading systems through massive requests or resource hogging.

Elevation of PrivilegeE-01

Privilege Escalation

Gaining privileges beyond what is intended.

Lateral MovementL-01

Lateral Movement

Accessing further internal systems after initial login.

Monitoring GapsM-AI-02

Overreliance on AI Outputs

Unchecked adoption of AI-generated content or code.

MITRE ATT&CK® Techniques

Reconnaissance

T1592

Gather Victim Digital Network Information

ATT&CK

Collecting information about the target infrastructure.

Mitigated by

Data Anonymization (Differential Privacy)Extended ProtectionHigh

NIST: PR.PT-3

Data Loss Prevention (DLP)Baseline ProtectionMedium

CIS: 13.3NIST: PR.DS-1

LLM Gateway / FirewallExtended ProtectionMedium

NIST: AI-1OWASP: LLM01

Continuous AI Red TeamingExtended ProtectionHigh

CIS: 18.1NIST: AI-5.1

Output Content FilteringBaseline ProtectionLow

NIST: AI-1.2OWASP: LLM02

Initial Access

T1078

Valid Accounts

ATT&CK

Exploiting existing credentials for access.

Mitigated by

Least Privilege PrincipleBaseline ProtectionMedium

CIS: 6.2NIST: PR.AC-6OWASP: A01:2021

Regular Access ReviewsBaseline ProtectionLow

CIS: 6.6NIST: PR.AC-1OWASP: A01:2021

Persistence

T1098

Account Manipulation

ATT&CK

Changing permissions or creating new accounts.

Mitigated by

Regular Access ReviewsBaseline ProtectionLow

CIS: 6.6NIST: PR.AC-1OWASP: A01:2021

Credential Access

T1552

Credentials from Password Stores

ATT&CK

Extracting passwords from web browsers or password managers.

Mitigated by

Output Content FilteringBaseline ProtectionLow

NIST: AI-1.2OWASP: LLM02

Collection

T1530

Data from Cloud Storage Object

ATT&CK

Accessing data from cloud storage (S3, Blobs).

Mitigated by

Data Anonymization (Differential Privacy)Extended ProtectionHigh

NIST: PR.PT-3

Data Loss Prevention (DLP)Baseline ProtectionMedium

CIS: 13.3NIST: PR.DS-1

Least Privilege PrincipleBaseline ProtectionMedium

CIS: 6.2NIST: PR.AC-6OWASP: A01:2021

T1213

Data from Information Repositories

ATT&CK

Accessing data from knowledge bases (SharePoint, Confluence).

Mitigated by

Data Loss Prevention (DLP)Baseline ProtectionMedium

CIS: 13.3NIST: PR.DS-1

LLM Gateway / FirewallExtended ProtectionMedium

NIST: AI-1OWASP: LLM01

Least Privilege PrincipleBaseline ProtectionMedium

CIS: 6.2NIST: PR.AC-6OWASP: A01:2021

Exfiltration

T1020

Automated Exfiltration

ATT&CK

Automated exfiltration of data via interfaces.

Mitigated by

Data Anonymization (Differential Privacy)Extended ProtectionHigh

NIST: PR.PT-3

LLM Gateway / FirewallExtended ProtectionMedium

NIST: AI-1OWASP: LLM01

Immutable Audit LogsExtended ProtectionMedium

CIS: 8.2NIST: PR.PT-1

Continuous AI Red TeamingExtended ProtectionHigh

CIS: 18.1NIST: AI-5.1

Output Content FilteringBaseline ProtectionLow

NIST: AI-1.2OWASP: LLM02

T1567

Exfiltration Over Web Service

ATT&CK

Data leakage via legitimate web interfaces.

Mitigated by

Data Loss Prevention (DLP)Baseline ProtectionMedium

CIS: 13.3NIST: PR.DS-1

LLM Gateway / FirewallExtended ProtectionMedium

NIST: AI-1OWASP: LLM01

SIEM IntegrationExtended ProtectionHigh

CIS: 8.5NIST: DE.AE-3NIST: DE.CM-1OWASP: A09:2021

Continuous AI Red TeamingExtended ProtectionHigh

CIS: 18.1NIST: AI-5.1

Impact

T1565

Inplace Modification

ATT&CK

Manipulation of existing code or data at the storage location.

Mitigated by

Data Anonymization (Differential Privacy)Extended ProtectionHigh

NIST: PR.PT-3

Immutable Audit LogsExtended ProtectionMedium

CIS: 8.2NIST: PR.PT-1

Is your scenario more complex?

AYSOLI experts support you in implementing your specific security requirements.

Free consultation