Meta is quietly testing a new Facebook feature that grants the platform access to users’ device photo libraries—including images and videos that have never been shared online. This development has raised serious privacy and transparency concerns among users and digital rights advocates.
Initially reported by TechCrunch, the feature appears as a pop-up prompt when some Facebook users attempt to upload a Story. It invites them to enable “cloud processing,” a capability that allows Meta to automatically scan and upload images from a user’s device gallery to its cloud servers. In return, users are offered AI-generated tools such as themed collages, memory recaps, and personalized filters for events like birthdays or holidays.
While the offering is positioned as a creative enhancement, opting in grants Meta sweeping access to a user’s camera roll. This includes metadata (such as timestamps and location), facial recognition data, and object detection—effectively transforming the user’s private gallery into a rich source of information for Meta’s artificial intelligence systems.
What’s particularly troubling is the rollout’s opacity. Meta has not issued a formal announcement or press release. Aside from a low-profile help page for Android and iOS users, the feature has emerged with little notice, leaving many users unaware of what they are consenting to. Once enabled, the upload process continues quietly in the background, raising concerns that private, unpublished content may be processed or analyzed without fully informed user consent.
Meta insists that the feature is optional and reversible. Users who disable cloud processing will reportedly have their unpublished media deleted from Meta’s servers within 30 days. However, the company has not definitively ruled out using this content for AI training in the future. Moreover, its updated AI Terms of Service—effective since June 23, 2024—do not clearly address whether data collected via this method is exempt from AI training applications.
This isn’t Meta’s first foray into large-scale data collection for AI. The company has previously acknowledged scraping public content from Facebook and Instagram to train its generative models. However, the boundaries between public and private data remain murky, especially as Meta increases its reliance on user-generated content to refine its AI tools.
The potential implications are even more significant in countries like India, where mobile devices often store sensitive information such as personal IDs, family photos, and confidential documents. Critics argue that the lack of localized explanations or language support may leave non-English-speaking users at heightened risk.
As Meta prepares for a possible global rollout, digital rights experts warn that the move could rekindle debates about algorithmic transparency, digital consent, and ethical data usage in the AI era.