Facebook has begun quietly testing a feature that allows Meta’s AI systems to access private photos stored on users’ phones — including images that were never uploaded or shared. Presented as a user-friendly enhancement for photo organization and editing, the new tool offers “cloud-based photo processing” capabilities inside the Facebook app. Once enabled, the feature allows Meta to analyze images for faces, objects, themes, and metadata directly from a user’s gallery. The rollout was first reported by G.Business, referencing a detailed breakdown published by PCWorld.

What the Feature Does — and Why It’s Risky

The system is currently in early-stage testing and is offered to users while creating Facebook Stories or editing media within the app. On the surface, the option appears to improve features like auto-collage creation or AI-based image enhancements.

But privacy experts are raising alarms: enabling the feature gives Meta's AI broad access to your entire photo library — not just shared content.“Even if images are stored offline and never uploaded, Meta can still scan them for pattern recognition, face mapping, and metadata,” notes a digital rights researcher in Berlin.

Meta’s Vague Response on AI Training

According to Meta, the tool is “opt-in” and not used to train AI models at this time. However, the company has not ruled out using such data for future machine learning purposes. “The lack of long-term guarantees is what makes this so concerning,” said a former data policy advisor in Brussels.
“Today it's convenience. Tomorrow it’s biometric profiling.”

Online, concerns are growing. Discussions on Reddit and X (formerly Twitter) show users expressing unease that private content could be indexed, profiled, or stored indefinitely.

TechCrunch

How to Disable the Feature

If you’ve already enabled the feature — or want to avoid doing so — here’s how to take back control:

  • Open the Facebook app
  • Go to Settings → Privacy → Preferences → Media Access
  • Toggle off the option for cloud photo processing
  • Recheck permissions in your phone’s system settings
  • Avoid uploading sensitive images you wouldn’t want scanned by AI

The feature is not mandatory, but consent may be presented during routine actions in the app, such as creating a Story or editing an image.

Broader Concerns About AI and Consent

Meta’s move comes amid industry-wide efforts to build AI models based on real-world user data. In the past, Meta (and other tech giants) have quietly expanded the scope of data collection — sometimes retroactively updating policies to reflect new uses.

“We’ve entered an era where the phone in your pocket isn’t just a device — it’s a training ground for AI,”
said an ethics researcher at the University of Edinburgh. Critics point out that Meta’s privacy language around the feature is broad and noncommittal, giving the company considerable legal leeway for future data usage.

Stay connected for news that works — timely, factual, and free from opinion. Learn more about this topic and related developments here: New Migration Rules: Poland Reinforces German and Lithuanian Borders