How Much Is Your Privacy Worth?

This week's piece departs from the usual territory. No product feeds. No agentic shopping. No search optimisation debates. Instead, something that sits upstream of all of those conversations — and one that will, sooner than most of us realise, reshape the relationship between technology, identity, and public life itself.

The subject is privacy. Or, more precisely, its accelerating disappearance.

On 20 July, New York becomes the first US state to impose a blanket ban on smart glasses across all 1,240 of its courthouses. The prohibition covers any eyewear or headwear containing cameras, microphones, or recording technology — prescription lenses included. Anyone arriving with a pair must surrender them to uniformed court officers before entering. Attorneys, staff, witnesses, members of the public. No exceptions.

The timing is not coincidental. It arrives in the same week that Meta announced it will now disable the camera on its smart glasses entirely if its system detects the LED recording indicator has been tampered with or destroyed. The update was necessary, Meta acknowledged, because users had been employing increasingly sophisticated methods to cover or destroy the light — enabling them to record people (often women) without any visible signal that they were being filmed.

These are not edge cases. They are symptoms of a structural shift that every person — not just every business — should be paying close attention to.

The "We're All Building the Same Thing" Problem

Meta has sold more than seven million pairs of its Ray-Ban smart glasses, making them some of the fastest-growing consumer electronics in recent history. EssilorLuxottica, which manufactures the frames, reported that sales more than tripled year-over-year in 2025. Meta holds an estimated 76 per cent of the global smart glasses market.

But Meta is not the only company pursuing this category. It is merely the furthest along. And every major technology company with consumer ambitions is converging on exactly the same thesis — that AI needs to move off the screen and onto the body.

Google and Samsung unveiled their jointly developed Gemini-powered smart glasses at Google I/O in May, with a launch confirmed for autumn 2026. The glasses run on Google's Android XR platform and will work with both Android and iOS devices — a cross-platform decision that signals serious intent to compete for mass adoption. Frames are being designed in partnership with Gentle Monster and Warby Parker, with luxury group Kering confirming it is developing Android XR glasses under the Gucci brand for 2027. (Because nothing says "surveillance capitalism" quite like doing it in Italian leather.)

Apple has reportedly redirected engineering resources away from its Vision Pro headset to prioritise smart glasses development. Codenamed N50, Apple's glasses are expected to ship in late 2027, with projected first-year volumes of three to five million units. The device will have no display — instead relying on cameras, microphones, speakers, and an upgraded Siri to deliver contextual AI assistance. A screenless Apple product. The irony of the company that made us all stare at screens now building a device specifically designed not to have one is genuinely delicious.

OpenAI acquired Jony Ive's hardware studio, io, for $6.4 billion in 2025 and has confirmed it will unveil its first consumer device in the second half of 2026. The device — described as screenless, voice-first, and potentially wearable — is being manufactured by Foxconn with an initial production target of 40 to 50 million units. Multiple form factors are reportedly under development, including an earbud-style wearable codenamed "Sweetpea" and a pen device codenamed "Gumdrop." All designed around the principle of always-on contextual awareness. (Nothing sinister about that phrasing at all.)

Amazon acquired Bee, a San Francisco startup making a $50 AI-powered wristband that passively records and transcribes conversations, in mid-2025. The device listens continuously — unless the user manually mutes it — with the stated goal of creating reminders, to-do lists, and personal insights from the wearer's daily life. Meta itself acquired Limitless, a maker of AI-powered pendants that can record and transcribe conversations in real time, in December 2025.

Qualcomm's CEO has projected that AI-equipped wearable devices could reach 100 million annual shipments within the next year or two, spanning glasses, earbuds with cameras, and even jewellery. Analysts at Smart Analytics Global project global AI smart glasses shipments alone could reach 75 million units by 2030.

So to recap: every major tech company is racing to build a device that straps sensors to your body and continuously observes the physical world around you. And they're all doing it simultaneously. This is not a trend. It is a coordinated species-level decision by Silicon Valley to turn human beings into mobile recording equipment.

The Privacy Architecture Is Being Built by the People Who Benefit from It Not Existing

The commercial logic is straightforward. An AI assistant that can see what you see, hear what you hear, and contextually understand your environment is dramatically more useful than one confined to a screen. Translation, navigation, object recognition, hands-free communication, contextual reminders — all of these applications require sensors that continuously observe.

The problem is that those sensors do not distinguish between the wearer and everyone else in range.

The Financial Times reported this week that Meta is testing a prototype of "super sensing" AI glasses that would continuously record audio while capturing photos every few seconds. The purpose is to create an always-on assistant that can recall what you saw, heard, or discussed throughout the day. According to multiple sources familiar with the project, executives are currently planning not to activate the LED recording indicator when these features are in use — making it impossible for bystanders to know they are being captured.

Meta's rationale, as laid out in a 2025 policy paper, is that if the LED blinked constantly during AI interactions, people would stop noticing it, thereby reducing its effectiveness as a signal during intentional recording. The logic is internally consistent. It is also the kind of reasoning that makes you want to read it twice to make sure you're not hallucinating.

The company has form. In June, WIRED discovered that Meta had already embedded substantial components of an unreleased facial recognition system — internally called "NameTag" — into its Meta AI companion app, which had been downloaded by more than 50 million users. The system included models for detecting faces, extracting biometric features, and matching them against stored profiles. It was inactive, but the infrastructure was live. Meta removed the code within a day of the report. (Which is the corporate equivalent of frantically shoving things under the sofa when guests arrive early.)

Earlier this year, workers in Kenya contracted to train Meta's AI using footage from smart glasses reported being required to view graphic content including sexual activity and bathroom footage captured unknowingly by users. Two lawsuits followed — one from people who had no idea such recordings existed, another from users who did not know their footage was being shared for human review.

Meta's response has been consistent: users are responsible for complying with applicable laws and using the glasses in a "safe, respectful manner." The company has teams dedicated to limiting misuse, a spokesperson told the BBC, "but as with any technology, the onus is ultimately on individual people to not actively exploit it."

That framing — positioning responsibility with the individual user while continuing to design hardware that makes misuse trivially easy — is doing a remarkable amount of heavy lifting.

The Consent Problem Nobody Has Solved (Because Solving It Would Be Expensive)

The legal landscape is fragmented and lagging. Several US states require all parties to a conversation to consent before it can be recorded. There are civil remedies for invasion of privacy, even in public spaces, including claims for disclosing private facts, false light depiction, and intrusion upon seclusion. But enforcement is sparse, and the law was not written with always-on wearable cameras in mind.

California has introduced a bill specifically targeting secret recordings with wearable devices. New York's courthouse ban is the most visible institutional response to date. Philadelphia courtrooms, some cruise ship operators, and the US Air Force have implemented their own restrictions. The University of San Francisco issued a public safety warning after a man wearing Ray-Ban Meta glasses was suspected of filming women on campus.

But these are piecemeal responses to a systemic problem. As one privacy law professor at Boston University put it, there is no single law that addresses the range of dangers these devices create. Individual institutions banning glasses from their premises is a bit like fighting a flood by putting sandbags in front of your own door — understandable, rational, and fundamentally not going to stop the water.

What We Are Actually Giving Up

The more fundamental question here is not a legal one. It is not even a commercial one. It is a question about the kind of world we are choosing to build.

For most of modern history, public life has allowed a degree of obscurity. You could walk into a café, sit on a train, browse a shop, have a conversation in a park — and reasonably expect that no permanent record of your presence was being created. A stranger might see your face, but they would not know your name, your employer, your political affiliations, or your dating history. That ambient anonymity was never a formal right in most jurisdictions, but it was a social reality that underpinned a certain kind of freedom — the freedom to exist in public without being observed, catalogued, and stored.

Smart glasses, at the scale now being projected, threaten to eliminate that reality entirely.

Consider what happens when tens of millions of people are wearing devices that continuously record their surroundings. Every coffee shop, every school run, every doctor's waiting room, every protest march, every moment of vulnerability or joy or boredom becomes potential footage — captured not by a surveillance state, but by the person sitting next to you. And when facial recognition is layered on top (as Meta has explored and as the underlying technology readily enables), a stranger can look at you and know who you are. Not in a vague, familiar sense. In a precise, searchable, cross-referenced sense — your name, your social profiles, your workplace, your connections.

Harvard students demonstrated exactly this in 2024, connecting Meta glasses to publicly available facial recognition databases to identify strangers in real time. What was then a research exercise is now, reportedly, on Meta's product roadmap.

The implications extend well beyond individual discomfort. A world without practical anonymity is a world in which behaviour changes. Research in surveillance studies has consistently shown that people self-censor when they know or suspect they are being watched. They become less likely to attend protests, less willing to seek sensitive medical advice, less open in conversation. The chilling effect is not hypothetical — it is one of the most well-documented consequences of surveillance, whether conducted by governments or enabled by commercial technology.

We've spent years talking about digital privacy in the context of cookies and ad tracking and data brokers. Those conversations suddenly feel quaint. The gap between "Meta knows what websites you visit" and "a stranger on the bus can identify you by looking at you" is not a gap. It is a canyon.

The Data Goes Somewhere. It Always Goes Somewhere.

And this is where the political dimension becomes impossible to ignore.

The data these devices generate does not simply evaporate. It flows somewhere — to servers, to training datasets, to platforms whose business models depend on knowing as much about human behaviour as possible. Meta is already discussing whether data collected through its glasses could be used to train its own AI models. The company trains its AI on images shared through Instagram unless users opt out. It has built features that analyse photos on users' camera rolls that have never been shared publicly.

The question of who holds that data, under what terms, and what they can do with it is not academic. Biometric databases, location histories, conversational transcripts, and behavioural profiles collected at population scale are not just advertising assets. They are political instruments. History offers no shortage of examples of how data collected for commercial purposes can be repurposed — by governments, by law enforcement, by political operatives — in ways that the original users never anticipated and would never have consented to.

When more than 75 advocacy groups wrote to Meta in April urging the company to abandon facial recognition for its glasses, their concern was not abstract. They specifically warned that the technology could be weaponised by stalkers, scammers, abusers, and government agencies to identify and track individuals — including immigrants, people of colour, and nonviolent protesters. US senators warned that the combination of Meta's data holdings and facial recognition capability makes the technology "uniquely dangerous."

This is not a problem confined to one company. It is structural. Every major technology firm now entering the wearable space is building devices predicated on the same exchange: unprecedented convenience for the wearer, delivered at the cost of involuntary exposure for everyone else. The wearer chooses to participate. The bystander does not. That asymmetry — one person's consumer choice becoming everyone else's surveillance reality — is the core of what makes this different from previous privacy debates.

The Question Worth Sitting With

The technology industry has a tendency to treat capability as inevitability — to assume that because something can be built, it will be built, and that because it will be built, it should be accepted. The smart glasses arms race is testing that assumption in unusually visible ways.

Former Meta AI researchers now predict this generation of glasses will face the same backlash that killed Google Glass more than a decade ago — though the difference this time is that the install base is already in the millions, not the thousands. The devices are cheaper, more stylish, and more capable. And the companies behind them are far more powerful than Google was in 2013. (When someone got punched in a San Francisco bar for wearing Google Glass, it was a news story. When seven million people are wearing Meta's version, punching is no longer a scalable solution.)

None of this means smart glasses will fail commercially. They may well succeed. But the terms on which they succeed — the privacy trade-offs society is asked to make, the regulatory frameworks that emerge, and the social norms that develop around always-on recording — will determine something far more significant than market share.

They will determine whether future generations inherit a world in which it is still possible to walk down a street without being identified, profiled, and filed. Whether public space remains a place where people can exist without performing for an invisible audience. Whether the right to be left alone — perhaps the most quietly important freedom any society can offer — survives the collision with a technology designed, at its core, to ensure that no one is ever left unobserved.

The Bottom Line

The devices are coming from every direction. Meta, Google, Apple, OpenAI, Amazon — all converging on the same thesis, all building variations of the same product, all powered by the same fundamental bargain: the wearer gets convenience, and everyone around them gets recorded.

This is not a debate about whether smart glasses will exist. They already do, in their millions. The debate is about whether we get any say in the terms — or whether we simply put them on and look the other way.

For those of us in the ecommerce world, the implications are significant but secondary. How AI wearables reshape product discovery, how visual search changes the purchase journey, how contextual commerce evolves — all of that matters, and we will cover it. But it matters less than the foundational question underneath: what kind of public life are we willing to accept?

We rarely get to see a societal trade-off this clearly before it happens. This is one of those rare moments. What we do with it — as consumers, as citizens, as people who still enjoy the basic luxury of walking into a room without being automatically identified — is genuinely up to us.

For now.

P.S. If you're wondering whether your phone already does most of this — it does, but it stays in your pocket. The difference between a phone and a pair of glasses is the difference between a security camera you can see and one hidden behind a painting. Same technology. Very different social contract.

P.P.S. Meta's official position remains that users are responsible for not misusing the hardware. Which is the same energy as a fireworks manufacturer saying "well, we put 'don't aim at people' on the box."

P.P.P.S. If you made it to the end of this one, I appreciate it. Normal service (platform drama, AI shopping chaos, the occasional sarcastic chart) resumes next week. But some things are worth taking seriously before they become the thing we all wish we'd taken seriously sooner.

— Jo

• Google opens Gemini's Nano Banana personalized image generation to all eligible US users free, previously locked to paid tiers. It pulls Gmail, Photos, YouTube, and Search data to render images matching your tastes without prompting.

• Google ships Nano Banana 2 Lite, its fastest, cheapest image model at $0.034 per 1K image and 4-second output, plus Gemini Omni Flash for video generation and conversational editing at $0.10 per second, both live for developers.

• X launches two hosted MCP servers letting Grok, Cursor, and Claude call the X API for full-archive search, user lookups, bookmarks, trends, and Article drafts, plus a docs server, all authed through the open-source xurl bridge.

• The em dash panic treats normal punctuation as an AI tell, pressuring writers to pre-defend their prose. The fix is not banning it but proving it earns its place: a useful dash turns the sentence, a lazy one just fogs it.

• A study of 21,559 US firms finds AI adopters grow headcount 10% over two years, but gains come almost entirely from high-intensity spenders; entry-level roles rise 12%, countering forecasts that AI triggers broad job loss.

📊 Sheet Savvy AI - Use AI to save hours on repetitive tasks in Google Sheets.

🗒️ Search+ - Chat with any PDF documents

📹 Creatify - Generate high-quality marketing videos from a simple product link or text description.

🧑 Taskade Custom AI Agents - Specialized tools designed to automate routine activities like research, task management, and content creation

😍It was great, all I needed

🙂Good, not great

🙄It sucked