A stranger points a camera at your face. You notice. You can walk away, object, or ask them to stop. That basic exchange has governed public recording for decades. Smart glasses deleted it.
Toluwa Omitowoju found that out the hard way. A man wearing Meta’s AI-powered glasses filmed her without asking. She didn’t know until friends — and strangers she hadn’t spoken to in years — sent her a clip that had racked up roughly 200,000 views across platforms.
“I had no say,” she told CBS News. Other women, she added, described worse: filmed by people they never identified, posted under burner accounts they’ll never trace back.
A Prank Format Built on an Engineering Blind Spot
The clip Omitowoju appeared in fits a format that’s exploded alongside wearable AI. Creators use smart glasses to film “pickup line” content, often without telling the person on camera. It’s cheap to produce. It performs well on Instagram and TikTok. And the subject has no way to spot it happening.
That’s the real story, and it isn’t really about Meta. It’s about what happens once a recording device stops looking like one. A phone raised toward someone’s face reads as a social signal. Glasses just read as glasses. Devices built to capture footage without anyone lifting a hand are exactly the point — hands-free capture is the pitch for hikers, technicians, and reviewers. It’s also what erases the one visual cue a bystander used to rely on to say no.
Adam Mosseri, the head of Instagram, addressed the problem directly in July. He said the platform would pull videos that harass people or invade their privacy, including pickup-line content specifically. That’s moderation after the fact. It does nothing for someone already filmed, uploaded, and viewed 200,000 times before anyone at Instagram sees the clip.
The Safety Feature That Didn’t Hold Up
Meta built a safeguard for exactly this scenario: an LED light that turns on when recording starts, backed by software meant to block filming if someone covers or disables that light. It’s the industry’s standard fix for “how do I know I’m being filmed?”
CBS News tested it. A reporter covered the light after recording had already started. The disabling mechanism didn’t stop the capture. Meta’s response narrowed the claim — tamper detection only blocks recording if the light gets covered before the wearer presses record, not after. The company says it keeps refining the technology.
That gap matters. A safeguard triggered only by pre-recording tampering stops sloppy mistakes, not deliberate concealment. Anyone who understands how it works can defeat it exactly where it counts: mid-recording, once a subject is already in frame.
The hardware has drawn scrutiny before this story broke. Texas Attorney General Ken Paxton opened an investigation into Ray-Ban Meta glasses in May, alleging the cameras and microphones could covertly capture people in private settings, including bathrooms. Meta’s statement at the time called privacy and data protection core to every product it builds.
The Legal Framework Wasn’t Built for This Either
A working LED wouldn’t fix the deeper problem anyway. Most U.S. states apply one-party consent rules to recording — only the person doing the filming needs to know it’s happening. That standard predates smartphones. It assumed recording devices were rare, visible, and deliberately operated.
AI glasses satisfy the letter of one-party consent while gutting the protection it used to provide. There’s no visible cue left for the other party to notice, object to, or step away from.
Omitowoju made the comparison herself: see a phone pointed your way, and you can step back or ask the person to stop. Smart glasses remove that option entirely.
Where This Leaves Wearable AI
Three systems are supposed to catch a problem like this. All three lag behind the hardware, which now ships across a market approaching $3 billion in 2026 alone.
- Platform policy acts after upload, once the view count already exists.
- Device safeguards catch pre-recording tampering, not concealment mid-capture.
- State consent law never required the subject’s awareness in the first place.
None of these gaps belong to Meta alone. Every company chasing this category — and the field is growing fast — hits the same three-layer failure unless it designs around the problem instead of patching it afterward. A brighter LED isn’t the fix. A recording indicator that fails safe when covered mid-capture, paired with platform detection that flags non-consensual formats before they reach six figures of views, would actually close the gap.
Until that redesign happens, AI glasses keep running an experiment on public consent with no opt-out for the people who never agreed to take part.
Related: AI Wearables Are Watching and Listening. Are We Ready?
