Why Did Google Lock Down the Google Photos API?

What changed in the Google Photos API?

Google's API updates page says the changes "take effect March 31, 2025". The Library API was not switched off. Its announcement says it "will no longer allow access to the elements in a user's library that were not uploaded by your app". An app can still manage what it uploaded itself, but it can no longer read the rest of your library, which is what backup, sync and migration tools relied on.

For reading photos an app did not upload, Google points developers to the Photos Picker API. Its guide to the picking experience says that after searching, "the user can select one or more items from the results", and that albums and favorites are not shown directly. You choose the photos by hand in a Google screen, and the app receives only those.

Why did Google lock down the API?

Google gives one reason: privacy. The announcement says the picker model "ensures both user privacy and data protection, while also safeguarding developers from inadvertently compromising user data", and that some Library API scopes were removed "to improve user privacy".

The cost of that choice falls on one group: people who want their whole library somewhere else. An app you authorise can no longer pull everything for you, so a full move now starts with an export you request from Google yourself.

If the question behind this is what Google does with the photos it keeps, that is a separate topic, covered in how Google Photos uses your pictures for AI training.

How does this affect you?

If you only use the Google Photos app and website, nothing changes. The change is about what other apps can read.

• Backup and sync - An app that mirrored your whole Google Photos library to a NAS or another cloud through the Library API can no longer read the photos it did not upload.
• Migration - Authorising a new service to pull your library is no longer what the API offers. A move now starts with a Google Takeout export, or with picking photos by hand.
• Photo management apps - An app that organised photos across several services sees only what you pick for it, or what it uploaded itself.

How do you get your photos out of Google Photos?

Google Takeout

Google's privacy policy says "you can export a copy of content in your Google Account if you want to back it up or use it with a service outside of Google". Go to takeout.google.com, select Google Photos, and request an export. What Google's Takeout help page says about it:

• It can take days - The process "could take from a few minutes to a few days", and Google emails a link when the archive is ready
• You choose the archive size - If your data is larger than the maximum size you pick, "multiple archives will be created"
• Dates stay in the files - Google says the photo metadata "still has the original timestamp preserved in the metadata embedded within the files". Metadata added in Google Photos, such as comments, goes to a separate JSON file
• Links expire - "Your archive expires in about 7 days." After that you request a new one

Putting the JSON metadata back into the files

If you want the extra metadata from those JSON files inside the images, the open-source GooglePhotosTakeoutHelper lists "Writing data from .json s back to EXIF data" among its features and works on a Takeout archive. EXIF is where most photo apps look for dates and GPS coordinates; the ExifTool documentation notes that GPS tags "are stored in a separate IFD within the EXIF information" of the image itself.

Manual download from the web

You can also select photos on photos.google.com and download them. That suits a few albums, not a whole library.

Which tools still get a library out of Google Photos?

On 26 October 2026 we read the own page of each tool this post names and recorded whether it still offers a way to get photos out of Google Photos, and how. Four of the five do, and two of those four work only on a Takeout export. Each value comes from the linked page, not from running the tool.

An earlier version of this post listed Photomyne as a migration service. Its FAQ describes scanning prints and saving scans to Google Photos, not the reverse. MultCloud's page says you can "choose photos, albums, videos, or your entire Google Photos library", but does not say which Google interface it uses since the Library API change.

Where to put your exported photos

• Another cloud service - Compare your options here. Viallo, iCloud, or Ente depending on your priorities
• A NAS - Synology with Synology Photos or Immich for self-hosted management
• External hard drive - No monthly fee. Keep two copies in case one drive fails

What should you look for in the next photo platform?

The API change did not delete anyone's photos, but it removed the one-step way to move them. These five properties decide how hard leaving the next platform will be:

• Data export - Can you download your whole library, and does the export keep dates and locations?
• Standard file formats - Are your photos stored as regular JPEG, PNG or HEIC files, or in a proprietary format?
• Documented access for other apps - Does the platform say what other apps can read, and has it changed that before?
• Metadata in the files - Are EXIF dates and GPS coordinates kept inside the files themselves?
• Transparent terms - Does the platform state what it does and does not do with your photos?

How Viallo handles this

Viallo stores photos at full resolution without compression. JPEG and PNG files are kept as the files you uploaded; HEIC files are converted to JPEG after their metadata is read. You can download an album as a zip file. There is no single archive of the whole library yet, so leaving means downloading album by album.

Viallo does not scan your photos with AI and does not use them to train AI models.

How was this checked?

On 26 October 2026 we fetched, with curl and Python's standard library, the own page of each of the five tools in the table, and the Google and ExifTool pages behind every statement about a third party, keeping the exact sentence that supports each value. No tool was run and no transfer was tried. Viallo's own behaviour was read from its source code. The results are in dataset.csv, and check.py fetches every page again and reports whether each supporting sentence is still there.

Frequently Asked Questions