Duplicates pile up from repeated downloads, copied project folders and backups. The safe way to find them compares contents, not names.
By MacMonitor · Published · Updated · macOS 26; app requires macOS 14 or later
Short answer: for photos, open Photos and use the Duplicates collection under Utilities. For other files, macOS has no built-in duplicate finder; compare file contents with a checksum in Terminal or use an app that hashes files, keeps one copy of each, and moves the rest to the Trash.
In the Photos app, open Utilities → Duplicates in the sidebar. Photos groups exact and near-identical copies and lets you Merge them, which keeps the best version and its metadata. Do not delete files from inside the Photos library in Finder.
Two files named report.pdf can be different, and report (1).pdf can be identical to final.pdf. Reliable duplicate finding compares a checksum of each file's contents. In Terminal:
# inspect checksums for files in Downloads (read-only)
find ~/Downloads -type f -size +1M -exec shasum -a 256 {} +Compare matching SHA-256 digests, then verify the paths and contents before removing a copy. This command only inspects files larger than 1 MB; it is not a complete duplicate finder. Filenames can contain spaces or newlines, so do not feed this output into an automatic deletion command. Keep at least one verified copy and a backup of irreplaceable files.
Start with a folder you understand, such as a collection of repeated downloads. Scanning every backup, archive and working project together creates a harder question: some matching copies are intentionally retained for recovery or delivery. Identical contents do not mean every copy has the same purpose.
A useful first pass has a clear scope and a clear keeper. If you are tidying Downloads, keep the copy with the location and name you want to use afterward. If you are reviewing a project, check whether its software expects a particular path. A content match does not guarantee that changing a path is harmless.
For each group, inspect the filenames, full paths and modification information. Open an unfamiliar file and confirm it is the document you expect. Keep a copy in a stable location that you can find again. Renaming a file is optional; deleting every member of a group is never part of a sensible duplicate cleanup.
There is also a difference between a working copy and a backup. Two copies on the same disk do not protect against that disk failing. At the same time, deleting an independent backup just because it matches the working file can remove protection against accidental edits. Decide the backup policy separately from the duplicate-finding exercise.
Imagine that the same PDF appears in Downloads, a client folder and an archive prepared for delivery. A checksum comparison establishes matching contents; it does not tell you which location matters. The Downloads copy may be redundant, the client-folder copy may be the one you work from, and the delivery archive may be a record you deliberately retain.
In this fictional example, removing only the Downloads copy is easier to explain than deleting two arbitrary matches. If the client folder is synced, also consider whether deletion propagates to another device. Write the keeper’s path down before acting when the folder structure is unfamiliar.
A document exported twice can look identical while carrying different metadata or internal formatting. Two photographs can depict the same scene without being identical files. A content-hashing workflow is designed to find exact matches; it is not a substitute for reviewing near-duplicates or deciding which photograph looks best.
For Photos, use its own duplicate-management tools and recovery controls. An image library is an application-managed package, so manually deleting files from its internal folders is a different and much riskier operation. For ordinary documents, compare contents and keep the result within a folder structure you understand.
A duplicate group’s apparent total can be larger than the storage actually recoverable. Hard links refer to the same underlying file, and APFS clones can share blocks. A cloud placeholder can describe a full document that is not downloaded locally. These differences are why a simple “number of duplicates multiplied by file size” estimate deserves caution.
Moving selected copies to Trash is also not the final reclamation step. Review that the keeper opens correctly and the expected folder still works. Only then decide whether to empty the relevant Trash items. Do not empty everything else in the Trash as an unnoticed side effect of reviewing one duplicate group.
Names help narrow a review but do not establish identical contents. Different revisions can share a name, and identical downloads can have numbered filenames. Use content comparison for exact duplicates, then make a separate decision about location and retention.
Behavior differs between tools. Downloading every cloud-only file can consume space and bandwidth before the comparison even begins. Check whether the tool skips placeholders and whether the original remains safe in the sync service before selecting anything.
No. For exact duplicates, contents match, while the preferred location may be older. For similar documents, a newer modification date does not establish that the newer file contains the desired revision. Open the document and judge its role.
Stop further cleanup. Inspect the Trash or the application’s recovery controls, and restore without overwriting another file blindly. If a backup is needed, follow its restore procedure and confirm the recovered document before deleting any remaining matches.
Choose a location that matches the team’s normal workflow and confirm ownership before removing shared copies. A content match does not decide whether a delivery archive, an approved revision folder or a collaborator’s working copy should disappear. Document the retained path and the reason for the choice when the collection matters to more than one person. Keep deletion and backup retention as separate decisions, and verify the retained file through the application used by the team. That small agreement can prevent a technically correct duplicate match from creating a confusing workflow change.
Duplicates scans a folder you choose, uses a quick size and first-64 KB check before a full SHA-256 comparison, and groups identical files. It always keeps one copy in each group and can never select that copy. Hard links are not counted as duplicates, iCloud-only files are skipped without being downloaded, and nothing inside your Photos library or app bundles is offered. The rest go to the Trash, with Put Back if you change your mind.