Technology

Google Photos to Label AI-Edited Images for Transparency

Published October 25, 2024

Google Photos has recently incorporated various artificial intelligence (AI) editing tools into its platform. Recognizing the significance of using these tools responsibly, the company is taking steps to ensure transparency regarding AI edits.

In a recent blog post, Google announced that it will include a specific tag indicating when photos have been edited with its AI tools, such as Zoom Enhance, Magic Eraser, and Magic Editor. According to Google, this initiative aligns with its commitment to responsible AI use, guided by its AI Principles.

A new label, "Edited with Google AI," will be displayed alongside the photo's metadata, which includes details like the file name, backup status, and camera information. However, this label will not appear directly on the photo itself. If users share the edited images on social media or through messaging, others will not be able to see this information unless they access the metadata. It’s worth noting that individuals aiming to conceal this information can easily strip the metadata from the images.

While Google Photos aims to enhance transparency, the visibility of the AI edit label within the platform may still require some effort from users, as accessing metadata isn’t a common practice for everyone. Nonetheless, social media platforms such as Facebook and Instagram could potentially utilize this metadata to create their own labels for such edits.

In addition to the AI-specific label, Google will also indicate when images are edited using non-AI tools like Best Take or Add Me through the International Press Telecommunications Council (IPTC) metadata.

John Fisher, the Engineering Director for Google Photos and Google One, mentioned that the initiative towards AI transparency is ongoing. Google is committed to gathering user feedback and exploring further methods to disclose AI edits clearly.

While the introduction of this label is a step forward in addressing the complexities introduced by AI in photography, it is not entirely foolproof. The effort seems more geared toward the creator of the image rather than the audience but marks a significant beginning in clarifying the lines that AI editing has obscured.

Google, Photos, AI, Transparency, Editing