Facebook has begun asking users for permission to access and process images stored on their phones, including those photos that have not yet been shared on the social media platform. The new feature appears when users create a “Story” on Facebook’s app, prompting them to opt into a “cloud processing” service. According to the pop-up displayed, this service offers creative suggestions, using AI to generate ideas such as collages, photo recaps, themed sets, or stylized edits.
When users select “Allow,” Facebook uploads images from their camera roll to its servers continuously, analyzing photo content including the time, location, themes, and even facial features. The company explicitly states that images uploaded in this way are private, not used for ad targeting, and viewable only by the individual user. However, by agreeing, users consent to Meta’s separate AI Terms, allowing the company to analyze uploaded media using its artificial intelligence technologies.
Meta’s AI policy allows the company extensive rights in what it may do with this data. According to these terms, Facebook’s parent company says that any images selected for processing—once shared with Meta—will be subject to analysis, modification, and could become the basis for newly generated AI-created content. Meta also retains the right to store and manage personal information gleaned from user interactions, image content, and other uploaded materials, to better inform and personalize AI-generated outputs. Additionally, the AI terms specify that human reviewers may examine interactions users have with the company’s AI systems.
While this feature does not seem to be new, Facebook users recently began noticing and commenting on the option, raising questions about its implications for privacy. Some have found Facebook automatically surfacing stylized images from older photos. Users on online forums have begun expressing confusion and concern about the scope of Meta’s AI policies, as well as discussing how to disable the feature within the app’s privacy settings.
The capability to access a user’s private photo archive provides Meta a substantial advantage as competition intensifies between major tech companies seeking to harness artificial intelligence. However, the inclusion of sensitive personal imagery—especially before it is intentionally uploaded to any social network—also presents significant privacy risks. As a result, this newest feature underscores an ongoing tension between innovative technological capabilities and user privacy concerns.
Meta has not responded publicly with comments clarifying this feature or addressing concerns voiced by users. Its current AI terms have been in effect since June 2024, but as of now, comparisons with past iterations are impossible, as Meta does not publicly archive previous versions on its site. The company had previously limited data training primarily to content users had already publicly shared on Facebook or Instagram, notably excluding data from pictures kept private in camera galleries. The expansion of their approach marks a notable change, raising new questions about transparency and user control.
