FT专题称AI智能眼镜销量快速增长但隐私反弹升温

AI is coming for your glasses
The young woman in hot-pink shorts is standing looking at her phone when the guy in glasses walks up to her. “You are so wow,” he says, and asks her on a date. “I just went through a break-up,” she replies. He asks if they can be friends, then tries to get her number. She smiles awkwardly and waves him away.
Soon after, the woman was shocked to discover that this brief encounter in a Target store had been secretly filmed by a man wearing a pair of Meta’s Ray-Ban camera glasses, after spotting herself in one of his “rizz videos” — clips of guys trying to pick up women.
“THIS is why these glasses or any version should NOT exist,” said Elyse Herrera, an esports broadcaster better known as Herculyse, who was recognised in the Instagram clip by a fan. “This is disturbing. Makes me sick physically, this is violating & WRONG,” she said in an X post.
The Instagram account that captured and posted the video is no longer accessible, suggesting it may have been one of a growing number that Meta is rushing to ban because they show women secretly filmed by men in smart glasses.
Social media companies’ whack-a-mole battle against abuse and trolls has been raging for years. But the arrival of smart glasses has opened the floodgates to a new form of subtle but intrusive surveillance, leading to bans on using them in British pubs and courtrooms.
Yet despite this summer’s backlash, executives at Meta, Google, OpenAI and other tech companies see this as just the beginning. They hope smart glasses and similar portable products could one day replace the smartphone as our main gateway to AI.
The big idea is that letting our virtual assistants see and hear exactly what we do will make our AIs richer and smarter, extending our own intelligence by boosting our memory and organising our lives. Soon, many in the tech industry predict, walking down the street whispering to the AI agents in our glasses will seem as normal as phoning a friend on AirPods.
“We think that AI is going to be really empowering for people,” says Alex Himel, vice-president of wearables at Meta. “And how are we going to use the AI? We think it’ll be wearable form factors.” Billions of people already wear glasses, he adds, and using these to talk to AI is easier than having to pick up a phone and open an app.
It is precisely this touted omnipresence that makes critics uneasy. “These glasses are cringe stalkerware and as we already see every cool person hates them,” said privacy campaigner Meredith Whittaker, president of the Signal Foundation that develops the eponymous secure messaging app. “They represent yet another Big Tech-enabled breach of the social contract.”
For all the privacy furore, smart glasses sales are growing rapidly. Analysts at Counterpoint Research estimate consumers will spend more than $1tn between 2026 and 2032 on wearable devices, as tech companies pack cameras, microphones and AI features into everything from earbuds to pendants.
Smart glasses alone will generate an estimated $44bn in sales by 2032, Counterpoint predicts. Tech companies from Google and Samsung to Apple are all preparing to launch digital eyewear of their own.
“Every time you reach into your pocket or your bag to pull [a phone] out, you’re taking yourself out of the moment that you’re in,” Himel says. “The reason we like wearables is because it’s always with you, it’s easy to use.”
Smart glasses will also give AI apps like Gemini, Siri and ChatGPT something they have lacked while locked inside a phone: eyes and ears. Meta’s vision, which also includes “super sensing” AI glasses that constantly record the world around the user, has only fanned privacy campaigners’ fears that smart glasses could soon become even more intrusive than viral rizz videos.
Some believe they will become tools capable of instant facial recognition and potentially other forms of AI-enabled surveillance, distributed to millions of people.
But the hype around the glasses has now reached a pitch that will be difficult to tamp down. “Tech companies have an incredible ability to push these tools out faster than society can meaningfully acclimate to the threat,” says Woodrow Hartzog, professor of law at Boston University School of Law. “By the time society catches up there’s already a degree of normalisation.”
‘Pervert glasses’
The activist group Everyone Hates Elon, which previously targeted Tesla founder Elon Musk, has been attempting to fight that normalisation by putting up subverted billboard ads parodying Meta’s own.
“The biggest advance in pervert technology since the trenchcoat,” reads one, while another showed the late sex offender Jeffrey Epstein wearing Meta’s Ray-Bans above the slogan: “Glasses for people who don’t do consent.”
“We will never have more money than billionaires but we can always be funnier and more interesting,” says an EHE co-founder, who doesn’t want to be identified by name.
They believe the backlash against the glasses is inseparable from wider social fears about the impact of AI on jobs and culture. “AI is inevitable, we aren’t saying don’t use it — we are saying use it for positive benefit for people. Don’t use it to use and abuse people.”
But campaigns against so-called pervert glasses have done little to dent the devices’ sales. Since the first iteration of Meta’s Ray-Ban-branded camera glasses launched as part of a partnership with eyewear giant EssilorLuxottica in 2021, analysts estimate that the companies have sold well in excess of 10mn units, including more than 7mn last year.
EssilorLuxottica said last month that AI glasses revenue had almost doubled year on year in its latest quarter. In June, Meta enlisted mega-influencer Kylie Jenner to design and promote its latest model.
Despite fears that the privacy backlash could start to affect sales of analogue eyewear brands, EssilorLuxottica said its partnership with Meta remains “solid”.
“While we’re not seeing a real impact on sales, we take the privacy concerns being raised seriously. Privacy has been a key consideration from the design stage,” it says. “As with any emerging technology, there is a period of adjustment.” The company adds that it is “working to establish the right standards” and to ensure consumers understand how to use the glasses responsibly.
Smart glasses sales are still small by smartphone standards; more than a billion phones are sold every year. But in a consumer electronics market that has been flattened by rising costs, due to components being redirected to AI data centres, smart glasses stand out as a bright spot of growth.
Meta dominates the market, with a roughly 85 per cent share of all smart glasses sold, according to George Jijiashvili of Omdia, a market researcher.
Even though today the devices are mainly used for listening to music and hands-free photos while parenting or doing sports, rather than chatting to AI, Jijiashvili predicts total smart glasses sales will grow quickly from 15mn units this year to 45mn by 2030.
“Meta borrowed the Ray-Ban brand awareness and the cool factor, which any tech company that’s not Apple struggles with,” Jijiashvili says, adding that EssilorLuxottica’s huge retail footprint also gave the devices a boost. “Meta wouldn’t have been able to achieve this kind of scale without this partnership.”
Meta’s success has become the envy of other tech companies, such as Google and Snapchat, that have long tried to turn camera glasses into a hit product.
A decade ago, a combination of privacy blowback and poor usability sank Google Glass, a more expensive and less capable forerunner to Meta’s Ray-Bans that is one of the $4tn internet giant’s more infamous failures.
Meta’s Himel says that the tech giant’s glasses have succeeded thanks to a series of refinements to the design, camera and speaker quality — and luck that its first version launched just a year before ChatGPT broke out.
“The thing that changed was when AI accelerated faster than we and the industry had anticipated, [we realised] these glasses that we’re already making are going to be amazing for AI,” he adds.
One former Meta executive involved in the product’s development says chief Mark Zuckerberg realised that if smart glasses were to sell, they would need to look good, so he went to EssilorLuxottica. The world’s largest eyewear company insisted that adding electronics should not make its glasses too heavy, despite pressure from Meta engineers to keep packing in more tech.
“There was a clash between the Franco-Italian style zealots and Silicon Valley engineering zealots — and the former won,” the executive says.
The Ray-Bans’ popularity has spurred other tech companies to seek out eyewear brands to partner with, says Michael Miller, founder of Digital Eyewear Labs, a start-up. Google and Samsung are working with Warby Parker and Gentle Monster on the first models of their AI glasses, which are due for release later this year.
“One of the unsolved problems is just the variety of styles” that consumers want from glasses, Miller says. “There are only so many people that Ray-Bans can fit.”
Apple is working on its own design for smart glasses that could be released as soon as next year, according to people familiar with the project. Leaks have suggested it is also looking to put cameras in AirPods.
However, hoping to preserve its hard-won reputation for protecting privacy, Apple may only allow its assistant Siri to access its wearable cameras to assist its AI responses, rather than allowing users to take normal photos with them.
Another hint at Apple’s plans for smart glasses came in January when it acquired Israeli start-up Q.AI. Patents filed by Q.AI show its technology being used in headphones or glasses, using “facial skin micro movements” to detect what people were saying even if no sound came out of their mouth. The tech could allow Apple users to whisper to Siri in public, offering a discreet alternative to talking out as growing numbers of AI users switch from keyboards to voice input.
Even Realities, a Chinese start-up that makes smart glasses with a small optical display, said that its devices have become popular with software developers who use them for hands-free programming. “You can have a coding terminal in front of you and just speak to it to code,” says CEO of Even Realities Will Wang. “You don’t have to be tethered to the desk.” Developers can talk to their AI agents through the glasses and the agents type back their responses in the glasses’ tiny display.
For more than a decade, both Meta and Apple have invested heavily in virtual and augmented reality headsets with much more sophisticated displays that can impose digital content on to whatever the wearer is looking at. But cost and technical barriers mean that those products, while still on the roadmap, are still some years away from reaching the mass market.
“The holy grail is the next set of glasses, the AR glasses, where you get a full display,” said Anthony Geffen, chief executive and creative director of Atlantic Productions, which has worked with several Big Tech companies to make content for VR and AR devices. “There is no doubt to me that glasses are the next big platform. But we are two or three years away from the beginning of that phase.”
‘You end up just using your phone’
To continue investing in that next generation, Meta needs to start generating significant revenues from its glasses. But for now, much of the income from its Ray-Bans goes to EssilorLuxottica.
The two companies’ deal means that each partner recognises revenue from the glasses it sells through its own retail channels. Given the Italian eyewear giant’s massive distribution network, it is booking the lion’s share of smart glasses revenue. That means Meta needs to convince Ray-Ban buyers also to become paying subscribers for its AI services.
Meta has said some AI glasses features will require a subscription to access fully, beyond certain usage limits. But it has put on hold a plan to charge users for extended access to “conversation focus”, which amplifies voices of people in front of the wearer in loud environments, after a backlash from customers who saw it as a paywall on a previously free feature.
Meta has also started to open up the device to third-party developers, with an eye on how Apple turned its App Store into a huge money-spinner.
“AI will become increasingly useful, increasingly personalised and increasingly agentic. The best assistant is not the smartest one, it’s the one that knows you the best,” Himel says. “You’ll see a trend to bring more and more of that into wearables.”
Meta is also testing “super sensing” glasses that would continuously collect audio and take photos every few seconds, acting as an AI-powered memory aid, people familiar with the matter told the FT last month.
The feature could be activated on existing glasses via a software update, without illuminating the indicator light that currently tells others that the device is recording, these people said. Meta declined to comment on “internal prototypes”.
Another potential feature that has provoked a firestorm of controversy even before its release is the ability to use AI glasses to recognise who the wearer is looking at. In June, Wired magazine uncovered code for a facial recognition system that was embedded in Meta’s smart glasses platform but unreleased. The company later removed the system.
Campaign groups including the American Civil Liberties Union and Electronic Frontier Foundation called the idea “dystopian” and warned it could be used to identify and stalk victims or “co-opted for police surveillance”.
Himel argues that such features could be “useful in specific contexts”, such as for blind people recognising friends in a restaurant.
“You can then think about misuses of similar technology,” he says. “And I think one of the things that we’re working very hard to do is figure out how to deliver tech in useful moments to the right communities while also preventing abuse.”
As it cracks down on misuse, Meta recently updated its glasses to disable the camera if it detects users trying to tamper with the LED that lights up when the Ray-Bans are filming.
Himel adds: “For the record, the ability to recognise any person in the world, whether it’s from a wearable or a phone or holding a laptop up to them, is a feature that literally nobody wants and is not something that we have interest in providing.”
For now, the reality of AI glasses is less dystopian than dysfunctional, according to Jijiashvili, who owns a pair of Ray-Ban Meta glasses.
“There are signs of this being very useful,” he says. “But everything takes a bit too long and it’s a bit too inconsistent, so you end up just using your phone.”
Not everyone in the tech industry is convinced that eyewear is the best way to access AI. Carl Pei, chief executive of London-based consumer electronics start-up Nothing, claims today’s smart glasses have “very high return rates and very low retention”. “I don’t see that changing anytime soon,” he says.
Nothing, which makes a range of smartphones and headphones, had been working on smart glasses of its own but shelved the project, Pei says. It continues to work on other AI-enabled devices.
“We just couldn’t find a really strong value proposition [for glasses] and I don’t think anybody else has found one yet,” he says. “The AI features are barely used at all. That will change as voice agents become more capable, but we’re not there yet.”