A company stores text files in Amazon S3 that contain customer chat messages, timestamps, and customer PII. The company must provide conversation samples to an external service provider for quality control. The provider needs to randomly select samples up to the most recent conversation. The company must not share customer PII. The solution must scale as conversations grow and have the least operational overhead. Which solution meets these requirements?

The correct solution leverages an S3 Object Lambda Access Point, which allows you to process data retrieved from S3 with an AWS Lambda function. When the external service provider accesses the Object Lambda Access Point, the associated Lambda function automatically redacts PII from the text files on the fly before the data is returned. This approach is scalable, has low operational overhead, and ensures PII is never shared. The incorrect option involving an EC2 instance creates a batch process that duplicates data and requires managing EC2 instances, increasing operational overhead. The DynamoDB option is unsuitable because the requirement is to provide conversation samples (text files), not structured non-PII data in a database. It also adds