Security, Compliance and Governance for AI Solutions practice questions

Attach a DynamoDB resource-based policy to the table allowing access from the SageMaker notebook

Why wrong: DynamoDB resource-based policies are not supported; IAM roles are required.

Store AWS access keys in the notebook and use those credentials to access DynamoDB

Why wrong: Embedding long-term credentials is insecure and violates best practices.

Launch the SageMaker notebook in a VPC with a security group that allows access to DynamoDB

Why wrong: VPC security groups control network traffic, not IAM permissions.

Enable Rekognition logging in the AWS Management Console

Why wrong: Rekognition does not have a separate logging feature; it uses CloudTrail.

Enable CloudTrail management events for the S3 bucket

Why wrong: Management events do not capture data plane operations like GetObject.

Enable S3 server access logs on the bucket

Why wrong: Server access logs capture HTTP requests but are not integrated with CloudTrail for auditing.

Use S3 server-side encryption with S3-managed keys (SSE-S3) and enable CloudTrail trail for S3 data events

Why wrong: SSE-S3 does not provide customer-controlled key management; CloudTrail alone does not cover transit encryption.

Use S3 client-side encryption and configure SageMaker to use HTTPS for inter-container traffic

Why wrong: Client-side encryption requires managing keys outside AWS; inter-container traffic encryption is not directly configurable by the customer.

Enable S3 default encryption with AES-256 and use AWS CloudTrail for S3 data events

Why wrong: Default encryption uses SSE-S3, lacking key control; transit encryption is not addressed.

{"Effect":"Allow","Action":["sagemaker:Start*","sagemaker:Stop*"],"Resource":"*"}

Why wrong: Too permissive; affects all notebook instances.

{"Effect":"Allow","Action":"sagemaker:*","Resource":"*"}

Why wrong: Allows all actions on all resources.

{"Effect":"Allow","Action":"sagemaker:*","Resource":"arn:aws:sagemaker:us-east-1:123456789012:notebook-instance/MyNotebook"}

Why wrong: Allows all actions, including deletion.

Encrypt the text using AWS KMS before sending it to Comprehend

Why wrong: Comprehend cannot process encrypted text; it requires plaintext input.

Use the AWS SDK with server-side encryption enabled for the API call

Why wrong: Server-side encryption applies to data at rest, not in transit; the API call payload is plaintext.

Create a VPC endpoint for Comprehend and send requests over the private network

Why wrong: VPC endpoint provides private connectivity but does not inherently encrypt the payload; HTTPS should be used.

Configure the SageMaker endpoint to use mutual TLS (mTLS) with client certificates

Why wrong: SageMaker endpoints do not support mTLS; VPC placement is the standard method.

Place the SageMaker endpoint behind an Amazon CloudFront distribution with an origin access identity

Why wrong: CloudFront does not enforce VPC origin; clients can access CloudFront from anywhere.

Use AWS STS to generate temporary credentials and require the client to sign requests with them

Why wrong: Signing requests does not restrict network origin; requests can come from anywhere.

Use S3 default encryption and enable Textract logging

Why wrong: Textract does not have its own logging; S3 default encryption uses SSE-S3.

Enable S3 server access logs and use S3 SSE-KMS

Why wrong: S3 server access logs do not provide the same level of auditing as CloudTrail data events.

Use S3 server-side encryption with S3-managed keys (SSE-S3) and enable S3 access logs

Why wrong: SSE-S3 does not use customer-managed keys; S3 access logs do not capture all API activity like CloudTrail.

Create an AWS CloudTrail trail to log all S3 data events

Why wrong: CloudTrail logs access, but does not provide encryption.

Enable encryption at rest for the SageMaker endpoint using the AWS Management Console

Why wrong: SageMaker endpoint encryption at rest applies to the ML storage volume, not the training data in S3.

Use AWS Security Token Service (STS) to generate temporary credentials for the notebook instance

Why wrong: STS provides credentials, but does not control internet access.

Enable CloudTrail data events for the Lex bot to capture conversation logs

Why wrong: CloudTrail does not capture Lex conversation logs; Lex uses CloudWatch Logs.

Enable S3 server access logs on the bucket storing the Lex bot configuration

Why wrong: Server access logs do not capture Lex bot execution logs.

Create an IAM role for Rekognition to assume and attach it to the Rekognition service

Why wrong: Rekognition cannot assume a role; the calling service must assume the role.

Configure the S3 bucket with a bucket ACL that grants read access to the Rekognition service

Why wrong: Bucket ACLs are not the recommended way; Rekognition uses IAM or bucket policies.

Attach a resource-based policy to the Rekognition collection allowing S3 access

Why wrong: Rekognition collections do not have resource-based policies.

The upload succeeds because S3 automatically encrypts objects at rest

Why wrong: Automatic encryption does not satisfy the condition requiring the header.

The upload fails because the policy does not allow PutObject without encryption

Why wrong: The policy allows PutObject with a condition; the failure is due to missing condition, not because encryption is required.

The upload succeeds because the bucket has default encryption enabled

Why wrong: The policy condition checks the request header, not the bucket's default encryption.

The Lambda function does not have an IAM role that allows Rekognition to call S3

Why wrong: Rekognition does not assume the Lambda role; it uses the bucket policy.

The bucket policy does not grant Rekognition access to the bucket

Why wrong: The policy grants s3:GetObject to Rekognition service principal.

The bucket policy does not include PutObject permission

Why wrong: Rekognition only needs GetObject to read images.

The ECS tasks are not configured to use a NAT gateway to reach the Bedrock endpoint

Why wrong: The VPC endpoint allows private connectivity without a NAT gateway.

The VPC endpoint policy does not grant the ECS task IAM role permission to invoke Bedrock

Why wrong: The policy allows all actions; the issue is network connectivity, not permissions.

The ECS task IAM role does not have the bedrock:InvokeModel permission

Why wrong: Missing permissions would result in an access denied error, not a timeout.

Use a SageMaker notebook instance with a lifecycle configuration to encrypt data with a customer managed KMS key, and restrict endpoint access using an IAM policy.

Why wrong: Incorrect: Notebook instance does not encrypt the endpoint or enforce key rotation; IAM policy does not prevent public endpoint access.

Enable S3 bucket encryption with SSE-S3, attach a bucket policy denying public access, and use an AWS Lambda function to rotate the S3 bucket key every year.

Why wrong: Incorrect: SSE-S3 does not allow customer-managed key rotation; bucket policy alone does not restrict endpoint access.

Apply SSE-KMS with an AWS managed key to the S3 bucket, and use a Lambda function to rotate the key every year. Disable public access to the endpoint using a VPC endpoint.

Why wrong: Incorrect: AWS managed key cannot be rotated by customer; VPC endpoint does not prevent internet access if endpoint is still publicly exposed.

Amazon Macie

Why wrong: Macie discovers and classifies sensitive data in S3, but does not redact model outputs.

Amazon SageMaker Model Monitor

Why wrong: Model Monitor detects data drift, not PII in model outputs.

AWS CloudTrail

Why wrong: CloudTrail logs API calls, not content filtering.

Use an S3 bucket policy that grants access to the root user of each target account.

Why wrong: Root user access is too broad; best practice is to use IAM roles.

Make the S3 bucket public.

Why wrong: Public bucket exposes data to the internet, which is insecure.

Use S3 cross-region replication to copy the artifacts to each target account's bucket.

Why wrong: Replication does not control access and may not comply with data residency requirements.

kms:ReEncrypt

Why wrong: Used to re-encrypt data under a different key, not for initial decryption.

kms:DescribeKey

Why wrong: Not necessary for decryption; used to view key metadata.

kms:GenerateDataKey

Why wrong: Used for creating new data keys, not for decryption.

AWS Config

Why wrong: Config evaluates resource configurations against rules, not real-time threats.

AWS Shield

Why wrong: Shield provides DDoS protection, not general threat detection.

Amazon Inspector

Why wrong: Inspector assesses vulnerabilities in EC2 and container workloads, not threat detection.

Store all data in encrypted DynamoDB tables and query from notebooks.

Why wrong: Not relevant; data must still be protected within Studio.

Use a VPC with no internet access and enable S3 encryption for all data.

Why wrong: Does not encrypt data in the Studio notebook environment itself.

Enable SageMaker Studio's default encryption using the AWS managed key for EFS.

Why wrong: Uses default key, not a customer managed key, which may not satisfy compliance.

Encode the prompts before sending to the model.

Why wrong: Encoding prompts does not prevent bypass; the model would decode and potentially still produce toxic content.

Use a different foundation model that is less susceptible.

Why wrong: Model susceptibility is not the root cause; guardrails can be bypassed regardless.

Restrict access to the model using IAM policies.

Why wrong: Access control does not prevent authorized users from submitting encoded prompts.