AI safeguards were modified

Synopsis

Field	Value
Activation Period	14 Days
Training Period	30 Days
Test Period	N/A (single event)
Deduplication Period	5 Days
Required Data	Requires one of the following data sources: AWS Audit Log OR Azure Audit Log
Detection Modules	Cloud
Detector Tags	Cloud AI Infrastructure Analytics
ATT&CK Tactic	Defense Impairment (TA0112)
ATT&CK Technique	Disable or Modify Tools (T1685)
Severity	Informational

Description

A cloud identity modified AI safeguards. MITRE ATLAS Technique: AML.T0015 - Evade ML Model.

Attacker's Goals

Evade model's restrictions and potentially accessing sensitive data or banned topics.

Investigative actions

• Examine changed safeguards filters and determine which AI models were affected.
• Investigate any unusual activity originating from the suspected identity.

Variations

Unusual modification of AI safeguards

Synopsis

Field	Value
ATT&CK Tactic	Defense Impairment (TA0112)
ATT&CK Technique	Disable or Modify Tools (T1685)
Severity	Low

Description

A cloud identity modified AI safeguards. MITRE ATLAS Technique: AML.T0015 - Evade ML Model.

Attacker's Goals

Evade model's restrictions and potentially accessing sensitive data or banned topics.

Investigative actions

• Examine changed safeguards filters and determine which AI models were affected.
• Investigate any unusual activity originating from the suspected identity.

AI safeguards were created or modified

Synopsis

Field	Value
ATT&CK Tactic	Defense Impairment (TA0112)
ATT&CK Technique	Disable or Modify Tools (T1685)
Severity	Informational

Description

A cloud identity created or modified AI safeguards.

Attacker's Goals

Evade model's restrictions and potentially accessing sensitive data or banned topics.

Investigative actions

• Examine changed safeguards filters and determine which AI models were affected.
• Investigate any unusual activity originating from the suspected identity.