My META AI glasses are making me question my own privacy

I bought a pair of Meta AI glasses recently. Hands down, they’re excellent-good enough that I’ve stopped thinking of them as a gadget and started thinking of them as an accessory. Which was rather the point.

One of my quieter resolutions this year has been to loosen my grip on my phone. Fewer reflexive pocket-checks, fewer screens unlocked for no reason. The glasses seemed like an elegant fix. Music, calls, questions, information, all without ever breaking stride. “Hey Meta, what am I looking at?” And it tells me.

Genuinely fantastic. Until my brain, ever the insufferable governance committee, raised its hand:

What exactly does this thing know about me? What else could it know? Have I simply strapped a very sophisticated data-collection device to my own face?

The more I read about AI-enabled wearables, the less comfortable I became with how casually I’d adopted one - not because I suspect my glasses of plotting anything, but because the relationship between technology, data, and privacy shifts considerably once the device leaves your hand and settles onto your face.

My phone knows what I search for. My glasses can know what I look at. My phone knows who I call. My glasses can hear who’s talking near me. My phone knows where I go, because I carry it. My glasses go everywhere I go, well, because I’m wearing them.

That distinction is the whole story.

It was never about one photo

The concern isn’t a single photograph or conversation. It’s the pattern: what I look at repeatedly, where I go, who I spend time with, what I ask, what catches my attention. Each data point is trivial. Assembled, they sketch a remarkably precise portrait of a person.

Which raises the question I hadn’t considered until now — whether I’ve made myself easier to profile without noticing.

We generally think of privacy as something surrendered deliberately: a name typed in, an email submitted, a box ticked, a photo uploaded. It feels transactional, we know the moment we’ve given something away. Wearable AI dispenses with that logic entirely. A great deal of this data isn’t volunteered. It’s simply observed while I go about my day.

The data nobody agreed to

And it isn’t always mine. I wear the glasses; someone else walks into frame. They haven’t installed an app, read a policy, or clicked “agree.” They were merely standing there. That is a fundamentally different privacy arrangement — one we haven’t remotely caught up to.

When AI stops recording and starts inferring

The part that unsettles me more: what happens once AI becomes proficient not just at recording, but at inferring? A photograph is a single data point. A collection of photographs, conversations, locations, and habits becomes a pattern — and patterns surface things we never explicitly disclosed.

This, to me, is the more interesting frontier of AI governance. We spend enormous energy worrying whether AI will produce the wrong answer. I’m increasingly preoccupied with the opposite risk: the right answer, inferred from information we never meant to share.

So, am I panicking?

Not particularly — and that’s the odd part. The glasses haven’t become essential. They’re useful, occasionally delightful, gradually woven into routine. But their disappearance tomorrow wouldn’t register as a crisis.

Which sharpens the actual question: if the benefit isn’t transformative, how much privacy am I willing to trade for it? A far harder question than whether the technology is “safe” — particularly given that most of us already live with a baseline level of exposure. Breaches happen. Platforms profile. Apps track. Much of it sits so far outside individual control that vigilance against every possible vulnerability starts to resemble unpaid, thankless overtime.

So: should I care? Yes. Should I be paranoid? No. Should I understand precisely what I’m consenting to? Without question.

Governance has to catch up

The real challenge was never going to be solved by telling people to stop using the technology. They won’t. Convenience, novelty, and seamless integration reliably beat caution. So governance must move at the pace of adoption, not years behind it — clearer rules on consent, retention, third-party sharing, inference, and security, with particular attention to the data belonging to people who never opted into anything at all.

The question privacy has asked for decades; what does technology know because I told it? — is being quietly replaced by a harder one: what does technology know simply because it was there with me?

That’s a governance conversation with considerable staying power.

The verdict

For now, I remain unresolved. Part of me suspects I’ve made my own profiling meaningfully easier. The other part is looking in the mirror, entirely convinced that Ray-Ban’s nerdy-professor aesthetic has done wonders for my sex appeal.

Governance can wait. The glasses are staying.