Keep redacted information out of Google Search

When publishing documents and images on the web, you may unintentionally publish information
beyond what is immediately visible to the human eye. In particular, information that you might
not see, or that was intended to be redacted, might be included in some document formats and
visible to search engines.

Because search engines index public material on the web, including images, content that is
not completely redacted can potentially be findable in search engines. Assistive technologies
like screen readers can make this seemingly “hidden” content more easily accessible, and
common image understanding techniques like optical character recognition (OCR) similarly make
it possible to search for this content.

Even though putting text in a tiny font, using a font color that’s the same as the background
the text is on, or covering text with an image may make something invisible to the human eye,
these methods don’t actually redact material in a way that prevents search engines from
indexing it and making it findable.

Similarly, some document types include information in various ways that aren’t immediately
visible. They might include the document’s change history, allowing users to see text that has
been redacted or altered. They might retain the full versions of images that contain cropped
or redacted information. There might also be metadata that’s included in a file, which is not
immediately visible, that may list the names of people who accessed or edited the file.

All of this information can remain even when a document is exported or converted from one
format to another. If you need to remove information from a file, it’s critical that the
information is removed completely from the file before that file is made public.

Here are some best practices for how to appropriately redact information from documents that
you don’t want to be indexed and made discoverable via Google Search.

Edit and export images before embedding them

Google Search lists images that it finds across the web, both those that are on web pages or
those that are embedded into various document formats. Embedded images are sometimes edited
using only the containing document’s editing tools. This can cause this redaction to fail when
an image is indexed apart from the document. That is why it’s best to edit images before
embedding them into a document, not after. In particular:

• Crop out unwanted information from images before embedding them into documents. Some
  document editing tools (such as word processors or slide creation tools) will maintain any
  uncropped images that you use in the public version of the document, so be sure to review
  the tool’s documentation thoroughly.
• Completely remove or obscure any text or other non-public parts of the image, as OCR systems
  may turn any image text seen into searchable text.
• Remove any undesired metadata.

After following the suggestions in this document, export or save the updated images as non-vector or
flattened image file formats such as PNG or WEBP. This prevents those parts of the images from
being inadvertently included in a public document.

Edit or remove unwanted text before moving to a public file format

Before you generate the public document, remove any text that you don’t want displayed in the
final version of the file. Move to a public format that does not keep your previous change
history. Here are more specific tips:

• Use proper document redacting tools if a file needs to have information redacted. For
  example, avoid placing black rectangles over text as a redaction method, as this can result
  in the text still being included in the public document.
• Double-check the document metadata in the public file.
• Follow the document redaction best practices
  for the format that you are using (PDF, image, etc).
• Consider information in the URL or file name itself. Even if a part of a website is
  blocked by robots.txt, the
  URLs may be indexed in search (without their content). Use hashes in URL parameters instead
  of email addresses or names.
• Consider using authentication to limit access to the redacted content. Serve the resulting
  login page with a
  noindex robots meta tag
  to block indexing.
• When publishing, make sure that the website is
  verified in Google Search Console.
  This allows quick removal action, if needed.

What to do if unredacted or improperly redacted documents are indexed in Search

1. Remove the live document from the website or location where you published it.
2. Use the Removals tool
   for the verified site to remove the documents in question from Search. Use a URL prefix if
   you need to remove many documents. For verified sites, a URL removal generally takes less
   than a day. This prevents the document in question from appearing for any searches for
   redacted content.
3. Host the properly redacted document under a different URL. This makes sure that any newly
   indexed version is of the new document, and not an older version of the document (since
   recrawling of URLs and updating them in a search index can take a bit of time). Update any
   links to those documents.
4. Contact any other site that may also be hosting the improperly redacted documents and ask
   them to take them down as well. Ask them to use the Removals tool in their Search Console
   account, or you can use the
   Outdated Content tool
   to ask Google’s systems to update the search results.
5. Allow the URL removal requests to expire (this happens after the URLs were either updated in
   the Google Search index, or after about 6 months).