What are the possible causes for this difference?

Explanations:

Amazon S3 does not use deduplication for objects. Each object is stored in its entirety, and S3 does not automatically reduce storage usage by eliminating duplicate data. Therefore, deduplication would not account for the difference in storage size.

If versioning is enabled on the S3 bucket, multiple versions of the same object can be stored. Each version takes up space, which can lead to the total size in S3 being higher than what is seen when selecting files in the console. This could easily explain the discrepancy in reported storage size.

Incomplete multipart uploads can consume storage space. When a multipart upload is initiated but not completed, the parts that were uploaded remain in S3 until they are explicitly deleted after a timeout period. This can lead to additional storage charges that may not be reflected in the visible file size.

Enabling AWS Key Management Service (KMS) for encryption does not increase the storage size of objects in S3. KMS encrypts the data at rest, but it does not add to the size of the objects stored in S3, so this would not cause a discrepancy in reported storage size.

Intelligent-Tiering optimizes costs by moving data between two access tiers (frequent and infrequent access) based on changing access patterns. While it may affect costs, it does not increase the actual storage size of objects in the bucket, so it cannot explain the discrepancy in storage size.