Facebook has introduced a new feature that enables its Meta AI to propose edits for photos stored on users’ devices, even if these images haven’t been shared on the platform. This functionality is now available to all users in the United States and Canada. By opting in, users can receive suggestions to post these AI-enhanced photos directly to their Facebook Feed and Stories.
Initially tested during the summer, this feature prompts users with a permission dialog box requesting access to allow cloud processing. This access allows Facebook to generate creative ideas from the user’s camera roll, such as collages, recaps, AI restyling, birthday themes, and more.
To facilitate these AI-driven suggestions, Facebook’s app uploads images from the user’s device to its cloud on an ongoing basis. Meta assures users that their media will not be used for ad targeting purposes and won’t be utilized to improve its AI systems unless the user edits or shares the photos on the social network. Users have the option to disable this feature at any time.
By agreeing to Meta’s AI Terms of Service, users permit their media and facial features to be analyzed by AI. The terms state that, by processing photos, Meta can summarize image contents, modify images, and generate new content based on the image. Additionally, Meta uses the date and presence of people or objects in photos to craft its creative ideas, providing the company with more information about users’ relationships and lives.
Granting Meta access to unshared photos could offer the company an advantage in the AI race by providing a wealth of user data, behavioral insights, and ideas for new AI features.
Settings for this feature are located under the Preferences section of Facebook’s Settings. On the Camera roll sharing suggestions page, there are two toggles: one allows Facebook to suggest photos from the camera roll when browsing the app, and the other enables or disables cloud processing, which lets Meta create AI images using camera roll photos.
Meta has been leveraging its position as a dominant social network to improve its AI technology. The company had previously announced it would train its image recognition AI on publicly shared data, including posts and comments on Facebook and Instagram. Last year, it also stated it would train its AI on images that Ray-Ban Meta users asked the device to analyze.
