Feeling seen
Would you embrace the world of watching and listening things or would you rather not?
Welcome, readers! Your weekly digest of digital surveillance and censorship news is here, and I’m your host, Anna Baydakova.
This week has reminded me how close we are getting to the point where we find ourselves surrounded by objects that can “see” and “hear” us without us noticing. I mean, even now, do we pay much attention to the surveillance cameras on the streets, in public transport, inside government facilities?
But that already familiar expectation of routine surveillance will feel different when cameras are not just mushroom-shaped plastic objects hanging from ceilings and poles; they are also robots making food deliveries and patrolling the streets, doorbell cameras and home devices, small, almost invisible wearable gadgets on people walking around, and whatever new shapes and forms may be invented soon.
How do our expectations of privacy change in a world like this? Do we get accustomed to feeling always seen? And not only seen – recorded in an easily retrievable, searchable form, in a digital archive of the physical life events that AI agents will be able to prowl through, like they are now prowling the internet?
Would you feel anxious? Would you behave differently? Would you restrain spontaneous impulses to do things that may look suspicious: lean in to smell a flower behind someone’s garden fence, check your hair in a parked car’s mirror, slap a sticker onto a light pole, etc.?
Or, on the contrary, would you feel safer knowing that whatever happens will be part of the record, so citizens will be “on their best behavior,” as Oracle CEO Larry Ellison said while talking about this future of total surveillance?
And while we’re thinking about that, some people are already taking the matter to court.
So let’s see what this week has brought us!
If you’re enjoying this newsletter, please take a moment to share it with your friends, colleagues, and social media followers.
Biometrics briefing
UK police will deploy AI facial recognition technology at the Appleby Horse Fair. – The Telegraph
Serbian police are using facial recognition technology by Japan’s NEC and Russia’s NTechLab. – Balkan Insight
Indonesia will enforce mandatory biometric registration for new SIM cards starting on July 1, 2026. – Antara
Ready to scan
Meta has already deployed the code for its smart glasses’ facial recognition feature, which the company has been planning to launch, Wired found out. According to the outlet’s analysis, AI models that will make the facial recognition feature possible are already in the Meta AI companion mobile app.
The app is connected to Meta’s smart glasses and is necessary to use its key features. While Meta has not officially announced the new feature, the code had been shipped to users’ phones no later than January, although it’s not live yet, according to Wired.
Meta’s plans to add facial recognition to its smart glasses, effectively turning them into wearable biometric surveillance devices, became known in February. The New York Times reported on a leaked corporate memo suggesting the launch of the feature in a “dynamic political environment where many civil society groups that we would expect to attack us would have their resources focused on other concerns.”
Soon after that, dozens of human rights organizations asked Mark Zuckerberg and Meta to cancel the plans due to privacy concerns. Texas Attorney General Ken Paxton even opened an investigation into the project.
Meta spokesperson Ryan Daniels told Wired that the company was only exploring the new feature, and the code “is just evidence of that exploration.” Meta has a troubled history with facial recognition technology. The company has paid over $2 billion in total to settle class action lawsuits over unlawfully collecting Facebook users’ biometric data in recent years.
However, it looks like the company never actually gave up on plans to launch facial recognition tools, Wired writes. Maybe, just wait until the political environment is “dynamic” enough.
Earbuds with eyes?..
In the meantime, another company is taking a completely different approach to “seeing” devices: Apple is in the final stages of testing its new airpods with built in cameras, as Bloomberg reported in May. The cameras will add to the Apple AI assistant’s capabilities: users can ask Siri about things they see while wearing airpods, the agent can notify them about important details of their surroundings, and more.
There still are hurdles to the plan: camera would eat up the battery capacity, Siri is not as advanced as the leading AI models on the market, plus, Apple’s approach to privacy would require a slew of precautions before releasing the product, Wired writes.
“For most intelligent AI devices, having vision is extremely important. But Apple is so privacy-conscious, and that’s been a big part of their marketing for quite some time now, that it’s a very difficult tightrope for them to balance,” Anshel Sag, principal analyst at Moor Insights & Strategy, told Wired.
Ring has been served
Those of you who read this newsletter regularly must remember the saga of Amazon slowly turning its Ring cameras – in fact, the entire network of Ring cameras – into a biometric surveillance dragnet. So now, someone is bringing this matter to court.
Charles Sigwalt, a Virginia resident, is asking the Washington Western District Court in Seattle to allow a class action against Amazon on behalf of all Americans who may unknowingly surrender their right to privacy by simply walking past a Ring camera, the complaint says.
Last December, Ring added an opt-in facial recognition feature to its cameras, named “Familiar Faces.” Two months earlier, the company also launched the “Search Party” function, allowing AI-powered search in footage from cameras in the same neighborhood – a feature that Ring advertised as helpful for looking for runaway dogs, but which was apparently built with humans in mind as well.
The company also rolled out the “Community Requests” feature in September, allowing law enforcement agencies to request footage from consenting users’ cameras. Hard not to see the beginnings of a perfect AI surveillance ecosystem – and experts immediately did.
While Ring users can opt in or out of the new features, people who walk into or past their homes can not do so – they may get their facial images captured and stored by Amazon without ever agreeing to that, Sigwalt says in his complaint:
“When Plaintiffs and Class members entered the homes and businesses of places which had Ring cameras that deployed Familiar Faces, they did not consent to have their privacy rights violated at the entrance way. However, that is exactly what happened.”
Amazon is gathering facial images of millions of people for its own profit, Sigwalt argues, while those caught on the Ring cameras are exposed to the risk of their digital identity being hacked and stolen from Amazon’s servers.
Sigwalt is seeking $5,000,000 in damages – the lower threshold for such class action lawsuits.
Ring’s facial recognition and search functions have caused outrage before: during the Super Bowl broadcast, viewers criticized the company’s ad dedicated to the “Search Party” feature, and in December, Massachusetts Senator Ed Markey questioned the Ring facial recognition capabilities as unwarranted surveillance.
They know what you will do this summer
Last week, we learned about China’s efforts to build predictive policing technology using AI-powered surveillance cameras that can analyze video footage in real time. This week, a new report by researchers at Vanderbilt University has told us about similar plans in the digital realm.
The researchers reviewed leaked documents from Geedge Network, the architect of China’s online censorship machine, which also exports the “Great Firewall” technology to neighboring countries, including Kazakhstan, Ethiopia, Pakistan, and Myanmar.
The new leak reveals details of Geedge’s work on teaching artificial intelligence models to analyze massive amounts of data related to people’s behavior online to build personal profiles and predict political dissent.
The firm wants to build behavioral profiles of users based on their location histories, social ties, browsing activity, and other metadata, linking physical movements to online activity, the report says.
Unlike traditional online censorship, this approach suggests that AI can predict undesirable behavior based on existing patterns and provide “continuous behavioral modeling across entire populations.”
“In systems like these, ordinary activities – where someone travels, whom they meet, what they read and say, what patterns they resemble – can become inputs into algorithmic judgments about the political risks they pose. Importantly, these systems move government intervention earlier in the chain. They identify the people and networks that could become a threat. They infer intent. They intervene before dissent has organized itself enough to be visible. The trigger for state action is no longer something the citizen did. It is something the state believes the citizen will do,” the report says.
According to the New York Times’ sources in the U.S. intelligence community, China currently doesn’t have access to enough advanced chips to actually build this kind of predictive technology. In particular, the Biden-era embargo on the newest chips slowed the project down, the Vanderbilt researchers wrote. However, those restrictions have been significantly rolled back by the current U.S. administration.
That’s how the AI surveillance supply chain may work: a U.S. company creates cutting-edge technology – a Chinese firm uses it to build an electronic “prophet” that infers what you are about to say and do – someone gets arrested for a thought crime in Kazakhstan.
Come the spy robots
In the meantime, autonomous surveillance devices accompanying humans and monitoring their behavior are closer than you may think. A Chicago TikToker had been waiting for his fiancée at the 7-Eleven parking lot when an AI video surveillance tower told him to stop loitering. The device is a mobile surveillance trailer built by a company named LiveView Technologies, according to Motor1.
The firm says that the tower uses AI to analyze video for “specific behaviors or postures that indicate potential criminal activity or life safety or security risks,” but that it doesn’t use biometrics and doesn’t record audio.
“For instance, the system looks for suspicious behaviors such as potential car break-ins, individuals assuming a catalytic converter theft posture underneath vehicles, pacing between parked cars, or an individual or vehicle lingering in a designated zone after hours or past a set threshold (e.g., a loitering timer). LVT customers control which behaviors, boundaries and timer thresholds the system follows, and they can easily modify them as needed,” the company told Motor1.
Being talked down to by a camera on a pole must feel much less pleasant than watching a robot dog patrolling the place – and the soccer fans in Texas soon will. Robotic dogs by Boston Dynamics and Hyundai are being deployed to watch for suspicious objects at the World Cup venues in Dallas, Chron reports.
According to Boston Dynamics, the robots don’t have facial recognition capabilities, but they do carry 360-degree cameras, can recognize sounds and identify thermal or chemical anomalies.
And on the other side of the world, Vietnamese police are testing drones that patrol the tourist districts of Hoan Kiem and Tay Ho. The drones broadcast video in real time to the Hanoi Police command center, as well as to local cops’ mobile devices, TechNode reports. The devices were developed by Vietnamese companies MiSmart JSC and Gtel Robot JSC.
Who knows, maybe the future where multiple policing drones roam around cities, watch people, issue warnings and fines, maybe even arrest violators – isn’t too far away.
***
And that’s all from me for this week, folks.
Stay vigilant!
Anna