Protecting AI workloads with GuardDuty
Amazon GuardDuty foundational threat detection and Lambda Protection helps you to better secure and detect threats to AI workloads built on AWS.
The foundational GuardDuty threat detection monitors AWS CloudTrail management events to detect suspicious and malicious activity in generative AI workloads created by using AWS services, including Amazon Bedrock and Amazon SageMaker. For example, GuardDuty can identify activities such as:
-
Unusual removal of Amazon Bedrock security guardrails
-
Change of model training data source that can potentially lead to data poisoning attack
-
Suspicious Amazon Bedrock model invocation
-
Unusual notebook instance or training job creation in SageMaker
-
Exfiltrated Amazon Elastic Compute Cloud credentials that may have been used to call APIs in Amazon Bedrock, Amazon SageMaker, or self-managed AI workloads on EC2 instances, EKS clusters, or ECS tasks.
GuardDuty Lambda Protection can help detect potential threats related Amazon Bedrock agents. This may include suspicious network activity such as cryptomining, and communicating with malicious command and control servers that can be caused by supply chain attack or complex prompting.
The following video shows how the associated findings would look.
