Smart glasses are causing havoc in India — and police wear them too
Consumer AI hardware is colliding hardest with the places that adopted it fastest: one story today about glasses that watch, one about a car that listens.
India is turning into the stress test for what happens when cameras you can't see go mainstream — and the state is on the wearing side. A Delhi designer who goes by Shubnam learned this the hard way: a rage-bait creator recorded them at a protest on Meta smart glasses styled like ordinary Wayfarers, and the mocking reel that followed — millions of views, a transphobic pile-on, AI-generated memes — is still online months later. "All it took was one unconsented video to rewind two decades of progress that I made all on my own," they told MIT Technology Review, which published the report today. The uglier finding is who else is filming. According to a petition before the Delhi High Court, police wore the same glasses to record student demonstrators continuously for weeks, including while they ate and rested; the solicitor general dismissed the case as "luxury litigation," and the students got ten criminal investigations instead.
The market is compounding the problem. Smart glasses are among the fastest-growing wearables in India according to IDC, and Reliance Jio is set to launch a sub-$105 rival to Meta's $420 pair by December — the sort of price that moves the device from gadget to default. Meta's answer to the safety question is a capture LED it says can't be turned off; a Wired investigation documented a market for "stealth mode" mods that kill the light, and the Financial Times reports internal prototypes that would continuously photograph and listen without triggering it at all. Meta told MIT Technology Review it plans to keep the indicator for active capture but not for AI features that interpret the wearer's surroundings — on the theory that a constantly blinking light would eventually stop being noticed.
The asymmetry is the story. A device the wearer opts into and everyone around them opts out of, sold cheap, with the police among the wearers and India's voyeurism laws written for a different era of abuse, means no crackdown is coming — Boston University law professor Woodrow Hartzog's point is that by the time society acclimates, the normalization has already happened. It is not an India-only problem; we recently covered Paris opens a criminal probe into smart glasses used to film women. India is just the first place where every pressure — price, consent norms, state adoption — points the same way at once.
Banma's AutoOmni 2.0 puts a 23-billion-parameter full-modal model in the car, off the cloud. Unveiled at the Yunqi conference in Hangzhou, the MoE model runs the whole cockpit — navigation, cabin controls, media, always listening without a wake word — and the company claims routine tasks match a cloud model ten times its size while complex tasks land at 80–90% of one. Treat the benchmarks as vendor numbers: they come from Banma's own release, with no independent testing. The adoption claim is the more interesting one — over 60% of automakers that put on-device AI in cars since AutoOmni's 2025 debut chose Banma's stack, the company says, and the Freelander 8 is the first production car running the model on Qualcomm's Snapdragon Cockpit Elite. CTO Si Luo's stated next step, EdgeOmni, points where the cockpit war is heading: from one smart car to a swarm of them sharing context.
What to watch: whether the Delhi High Court does anything with the police-glasses petition, and whether JioFrames' budget launch detonates the consent debate or ends it.
Would you accept a camera you can't detect in a crowd — if the wearer paid for it, and so did the police? Tell us in the comments.
