In April of this year, a former executive of Facebook (Meta) accused the management team of Facebook of transferring the data of Americans, Taiwanese, and Hong Kongers to the Chinese Communist Party. Recently, a cybersecurity expert issued a warning that Meta may be using AI to scan users’ mobile photo albums to generate new content without the full understanding of the users.
Facebook recently introduced a new feature where users creating new Stories on the social networking app will be prompted to use an AI tool to edit photos. When users engage in this feature, a screen will appear asking if they choose “Cloud Processing” to allow for creative suggestions.
According to Facebook officials, choosing “Cloud Processing” means that Facebook can generate new creative content from the photos taken on your phone, making photo sharing more fun. Examples of such content include collages, photo reviews, AI style reshaping, or themed photos.
Facebook also mentioned that this feature will continuously send photos and videos from your phone’s album to their cloud services based on information such as time, location, or themes, even if these have not been uploaded to Facebook. While the message claims that “only the user can see these suggestions, and the media will not be used for ad targeting,” clicking “Allow” implies agreeing to Meta’s AI service terms.
The terms state that “once shared, the user agrees to Meta using AI to analyze and edit photos or videos from the phone, including facial features and content in the photos. These contents will be used for AI training and generation.” Additionally, Meta will use timestamps, the presence of individuals in photos, to generate so-called “creative content.”
This move has led to speculation that Meta’s existing public data may not suffice for its AI training needs. Previously, Meta announced using publicly shared posts and comments on Facebook and Instagram to train AI, but only EU users have the right to decide whether their public data can be used for AI training by May 27, 2025.
A spokesperson for Meta, Maria Cubeta, confirmed that this new feature is still in the testing phase, aiming to allow users to “easily share content,” and currently testing to recommend sharable and curated content from users’ phone albums.
Cubeta explained that “these suggestion features are limited to select users who opted in, and the content will only be shown to the user (unless decided to share), and can be turned off at any time.” Cubeta also emphasized that “media from the phone may be used to improve these suggestions but will not be used to improve the AI models in this current test.”
However, Meta’s new approach may not convince users and could even lead to new “slippery slope” effects.
The main reason is that like other tech giants, Meta also has ambitious goals with AI and aims to acquire more undisclosed photos and data to train its AI, striving to be a winner in the AI competition.
This new feature by Meta may not clearly explain the terms to users, which has raised concerns.
In early July, Caitlin Sarian, a cybersecurity expert with over 1.11 million followers on Instagram, issued a warning. She shared a post on Instagram, educating people on how to protect their media and privacy from being used by Facebook.
She advised, “Go to the ‘Menu’ on the bottom right of the Facebook app, scroll down to ‘Settings & Privacy,’ and click on ‘Settings.’ Then, in the search bar at the top, type in ‘Camera Roll Sharing Suggestions.’ Make sure that ‘Get suggestions for creating content when browsing Facebook’ and ‘Enable cloud processing of your album to get personalized creative content’ are both turned off.”
She continued, “Honestly, I would go further and outright restrict Facebook’s access to photos. To restrict access, click on ‘Apps,’ then ‘Facebook,’ ‘Photos,’ and choose one of the options – none, limited, or full access. I chose ‘limited’ access.”
Choosing “none” or “limited access” will restrict Facebook from accessing photos on your device without further permission.