AI Defense Administration
Provide the details of your organization's AI-powered applications and their associated AI model endpoints to enable seamless integration within the system. By entering these details, you establish a connection between your applications and the AI Defense platform, allowing for effective monitoring, compliance, and protection.
These integrations streamline the interaction between diverse systems and applications, ensuring smooth data flow and enriched functionality across platforms. These integrations enable consistent application of policies, guardrail monitoring, and runtime protection, allowing your organization to maintain compliance and mitigate risks associated with AI-powered operations. Additionally, by linking model endpoints, you can track usage patterns, assess risks, and receive actionable insights tailored to your organization's unique needs.
Connect Multicloud Defense
You must connect to Cisco Multicloud Defense before you can use the AI Assets feature to detect AI workloads in your environment.
To connect AI Defense to Multicloud Defense, you must provide your:
-
Multicloud Defense account name
-
Multicloud Defense API key
-
Multicloud Defense API secret
Connect Secure Access
You must connect to Cisco Secure Access before you can use the AI Access feature to show third-party AI applications that your users connect to.
To connect AI Defense to Secure Access, you must provide your:
-
Secure Access account name
-
Secure Access API key
-
Secure Access API secret
Connect AWS Bedrock
You must connect to AWS Bedrock in order to scan AI models and applications hosted there.
To connect:
-
Find or create an IAM role with permission to invoke models in AWS Bedrock. Make sure this role has access to all models you plan to scan. Copy the role ARN.
-
In AI Defense, navigate to Administration, Integrations, AWS Bedrock. Click Edit.
-
In the Connect AWS Bedrock pane, Paste the role ARN into the Bedrock Inference IAM Role ARN field and click Connect at the bottom of the pane.