Zuckerberg’s AI glasses ‘spy on people on the toilet’

Zuckerberg’s AI glasses ‘spy on people on the toilet’
Chief Lizard

In the dim glow of a Nairobi office, a young data annotator named Aisha (not her real name) stares at her screen, heart sinking. She’s just watched a stranger halfway across the world step naked from a shower, glasses left recording on the bedside table. Another clip shows a couple mid-intimacy; a third, someone fumbling with bank cards in the bathroom. “We see everything,” she whispers to investigators later, voice trembling. “From living rooms to naked bodies… I don’t think they know, because if they knew they wouldn’t be recording.”

This is the human face of Meta’s Ray-Ban AI smart glasses scandal. Launched with slick marketing—stylish frames promising hands-free real-time AI video processing—the glasses let users say “Hey Meta” and instantly query the world around them. The camera and mic spring to life, streaming live footage to Meta’s cloud for instant answers about objects, text, or scenes. Convenience at your fingertips. Or so it seemed.But behind the cool tech lies a hidden pipeline. When AI features activate, captured video isn’t confined to your phone. To train and refine the models, Meta routes some footage to human reviewers at subcontractor Sama in Kenya. Thousands of underpaid annotators there label objects, transcribe chats, and flag issues, ordinary work that has become anything but. Workers describe sex scenes filmed mid-act, toilet visits, partners undressing, even porn viewed through the wearer’s eyes. Faces sometimes remain visible despite promised blurring; lighting glitches defeat the filters. “You understand that it is someone’s private life you are looking at,” one said, “but you are not supposed to question it. If you start asking questions, you are gone.”

For glasses owners, busy parents, travelers, professionals, the invasion feels visceral. You slip on the shades for a quick recipe lookup in your kitchen, never imagining strangers in Africa are glimpsing your bedroom, your body, your most unguarded moments. The subtle recording light offers little warning during live AI sessions. Users trusted Meta’s privacy promises, yet the data flows globally regardless. Worse, Meta has quietly moved the goalposts on consent. Recent policy tweaks keep AI camera features enabled by default; turning them off requires deliberate steps. Opt-outs for cloud voice storage vanished. Buried in the AI Terms of Service lies the fine print: “In some cases Meta will review your interactions with AIs… and this review may be automated or manual (human).”

To unlock the glasses’ headline features, you must agree. What began as “optional sharing for better experiences” evolved into mandatory data surrender. Meta insists media “stays on the user’s device” unless shared and is filtered first, but Kenyan workers and the Swedish investigation prove otherwise. The backlash is swift and human. Europe’s GDPR watchdogs and the UK’s Information Commissioner’s Office are demanding answers, calling the revelations “concerning.” A U.S. class-action lawsuit accuses Meta of shipping intimate footage overseas without clear disclosure.

For Aisha and her colleagues, it’s ethical whiplash; peering into lives they’ll never meet, all to polish an AI that profits Meta while they earn scraps. This isn’t dystopian fiction, it’s the cost of wearable AI. A mother in London adjusts her glasses while helping her child dress, unaware her private space is now data in Nairobi. A couple in New York whispers goodnight, their tenderness labeled for training. Meta’s hunger for ever-smarter models has turned personal real-time video into a global privacy nightmare. Users deserve better than buried clauses and shifting rules. Until Meta stops treating our lives as training fodder, the glasses’ shine feels less like innovation and more like surveillance by another name. Check your settings. Think twice before saying “Hey Meta.” Your most intimate moments may already be someone else’s workday.