On Friday, Meta announced that the AI photo editing suggestions feature for Facebook is now available to all users in the United States and Canada. This feature allows Meta AI to access photos in users' phone camera rolls that have not yet been shared and provide editing suggestions, encouraging users to post these AI-edited photos on their Facebook feeds and Stories.
This feature was first tested this summer. When users open the Facebook app, a permission dialog box pops up, requesting "Allow cloud processing" so users can "get customized creative ideas from your camera roll." The prompt explains that the feature may provide creative suggestions such as collages, retrospectives, AI-style reworkings, birthday themes, etc.
From a technical perspective, for the AI to work, the Facebook app needs to continuously upload images from users' devices to its cloud. This allows Meta AI to generate editing suggestions. Meta states that users' media content will not be used for ad targeting or improving the AI system unless users edit the media or share the edited photos with friends or others on social networks.
This feature can be disabled at any time. Users can find the relevant option in the "Preferences" section of Facebook settings. On the "Camera Roll Share Suggestions" page, there are two switches. The first allows Facebook to suggest photos from the camera roll while browsing the app, and the second is used to enable or disable "cloud processing," which allows Meta to use photos from the camera roll to create AI images.
From a privacy perspective, although Meta claims it does not train AI on all photos, when users agree to Meta's AI service terms, they allow the AI to analyze their media content and facial features. The terms state that by processing photos, Meta has the ability to "summarize image content, modify images, and generate new content based on images."
Meta also uses dates, people, or objects in photos to develop creative ideas, providing Meta with more information about users themselves, their relationships, and their lives. Additionally, allowing Meta to access photos users have not shared on their platform could give the company an advantage in the AI race by providing large amounts of user data, behavioral insights, and new AI feature ideas.
From a strategic perspective, this feature is another move by Meta to leverage its social network dominance to improve AI technology. Meta has previously announced plans to use publicly shared data (including posts and comments on Facebook and Instagram) to train its image recognition AI. EU users have until May 27, 2025, to opt out. Last year, Meta also stated that it would use images analyzed by Ray-Ban Meta users' devices to train AI.
The launch of this feature has raised concerns about privacy and data usage. Although Meta provides options to opt out and statements about usage limitations, uploading unshared photos to the cloud for AI processing essentially expands Meta's access to users' private data. Users' camera rolls typically contain a large number of personal photos, including family, friends, private events, and daily life details, which hold far greater value than content already shared on social platforms.
From a business perspective, such features help Meta achieve multiple goals: increasing the amount of content users post on the platform, boosting user engagement, and accumulating richer training data for its AI system. Even if Meta claims it will not use unedited photos for AI training, the analysis of photo content, people, and scenes itself provides valuable user behavior and preference data.
Users need to weigh convenience against privacy protection when deciding whether to enable this feature. While AI editing suggestions may bring interesting creative effects, the cost is exposing their private photo library to Meta's cloud system. Considering Meta's history with data privacy and regulatory disputes, cautious users may choose to keep this feature turned off.
