Cameras, Cameras Everywhere!
We live in an age when a single walk down the street can put you inside at least a dozen different recording ecosystems at once: Fixed municipal CCTV, a bypassing police cruiser’s cameras or body-cam feeds, the license-plate cameras on light poles, the dash-, cabin-, and exterior cameras of nearby cloud-connected vehicles, Ring and Nest doorbells of residences that you might pass by, and the phones and wearables of other pedestrians passing you, that are quietly recording audio and/or video. Each of those systems was justified as a modest safety, convenience, or product feature, yet when stitched together they form a surveillance fabric that reaches far beyond its original intent. Instead of only looking at the big picture all these individual systems paint, let’s instead focus on each individual area and uncover some of the actors complicit in the making of this very surveillance machinery that they profit immensely from. Note: The lists below only mention a few of the most prominent enablers and profiteurs. CCTV is not new, but it’s booming. Market reports show the global video-surveillance/CCTV market measured in tens of billions of dollars and growing rapidly as governments and businesses deploy these solutions. A continued double-digit market growth over the next several years is expected. Cameras haven’t been reliably proven to reduce crime at scale, and the combination of live feeds, long-term storage and automated analytics (including behavior detection and face matching) enable discriminatory policing and concentrate a huge trove of intimate data without adequate oversight. Civil liberties groups and scholars argue CCTV expansion is often implemented with weak limits on access, retention, and third-party sharing. In addition, whenever tragedy strikes it seems like “more video surveillance, now powered by AI” is always the first response: More CCTV to be installed in train stations after knife attack Heidi Alexander has announced that the Government will invest in “improved” CCTV systems across the network, and that facial recognition could be introduced in stations following Saturday’s attack. “We are investing in improved CCTV in stations and the Home Office will soon be launching a consultation on more facial recognition technology which could be deployed in stations as well. So we take the safety of the travelling public incredibly seriously.” Automatic license-plate readers (ALPRs) used to be a tool for parking enforcement and specific investigations, but firms like Flock Safety have taken ALPRs into a new phase by offering cloud-hosted, networked plate-reading systems to neighborhoods, municipalities and private groups. The result is a searchable movement history for any car observed by the network. Supporters point to solved car thefts and missing-person leads. However, clearly these systems amount to distributed mass surveillance, with weak governance and potential for mission creep (including law-enforcement or immigration enforcement access). The ACLU and other groups have documented this tension and pressed for limits. Additionally there has been a plethora of media frenzy on specifically Flock Safety’s products and their reliability : A retired veteran named Lee Schmidt wanted to know how often Norfolk, Virginia’s 176 Flock Safety automated license-plate-reader cameras were tracking him. The answer, according to a U.S. District Court lawsuit filed in September, was more than four times a day, or 526 times from mid-February to early July. No, there’s no warrant out for Schmidt’s arrest, nor is there a warrant for Schmidt’s co-plaintiff, Crystal Arrington, whom the system tagged 849 times in roughly the same period. ( via Jalopnik ) Police departments now carry many more mobile recording tools than a decade ago, that allow the city’s static CCTV to be extended dynamically: Vehicle dash cameras, body-worn cameras (BWCs), and in some places live-streaming CCTV or automated alerts pushed to officers’ phones. Bodycams were originally promoted as accountability tools, and they have provided useful evidence, but they also create new data flows that can be fused with other systems (license-plate databases, facial-recognition engines, location logs), multiplying privacy and misuse risks. Many researchers, advocacy groups and watchdogs warn that pairing BWCs with facial recognition or AI analytics can make ubiquitous identification possible, and that policies and safeguards are lagging . Recent reporting has uncovered operations where real-time facial-recognition systems were used in ways not disclosed to local legislatures or the public, demonstrating how rapidly policy gets outpaced by deployment. One of many recent examples consists of an extended secret live-face-matching program in New Orleans that led to arrests and subsequent controversy about legality and oversight. Drones and aerial systems add another layer. Airborne or rooftop cameras can rapidly expand coverage areas and make “seeing everything” more practical, with similar debates about oversight, warranting, and civil-liberties protections. Modern cars increasingly ship with external and internal cameras, radar, microphones and cloud connections. Tesla specifically has been a headline example where in-car and exterior cameras record for features like Sentry Mode, Autopilot/FSD development, and safety investigations. Reporting has shown that internal videos captured by cars have, on multiple occasions, been accessed by company personnel and shared outside expected channels, sparking alarm about how that sensitive footage is handled. Videos of private interiors, garages and accidents have leaked, and workers have admitted to circulating clips . Regulators, privacy groups and media have flagged the risks of always-on vehicle cameras whose footage can be used beyond owners’ expectations. Automakers and suppliers are rapidly adding cameras for driver monitoring, ADAS (advanced driver-assistance systems), and event recording, which raises questions about consent when cars record passengers, passers-by, or are subject to remote access by manufacturers, insurers or law enforcement, especially with cloud-connected vehicles. Ring doorbells and other cloud-connected home security cameras have created an informal, semi-public surveillance layer. Millions of privately owned cameras facing streets and porches that can be searched, shared, and, in many jurisdictions, accessed by police via relationships or tools. Amazon’s Ring drew intense scrutiny for police partnerships and for security practices that at times exposed footage to unauthorized access. A private company mediates a vast public-facing camera network, and incentives push toward more sharing, not less. Another recent example of creeping features, Ring’s “Search Party” AI pet-finder feature (enabled by default), also raised fresh concerns about consent and the expansion of automated scanning on users’ cloud footage. While smartphones don’t (yet) record video all by themselves, the idea that our phones and earbuds “listen” only when we ask them has been punctured repeatedly. Investigations disclosed that contractors for Apple, Google and Amazon listened to small samples of voice-assistant recordings, often including accidentally captured private conversations, to train and improve models. There have also been appalling edge cases, like smart speakers accidentally sending recordings to contacts, or assistants waking and recording without clear triggers. These incidents underline how easily ambient audio can become recorded, labeled and routed into human or machine review. With AI assistants (Siri, Gemini, etc.) integrated on phones and wearables, for which processing often requires sending audio or text to the cloud, new features make it even harder for users to keep control of what’s retained, analyzed, or used to personalize models. A recent crop of AI wearables, like Humane ’s AI Pin , the Friend AI pendants and similar always-listening companions, aim to deliver an AI interface that’s untethered from a phone. They typically depend on continuous audio capture and sometimes even outward-facing cameras for vision features. The devices sparked two predictable controversies: Humane ’s AI Pin drew mixed reviews, questions about “trust lights” and bystander notice, and eventually a shutdown/asset sale that stranded some buyers, which is yet another example of how the technology and business models create risks for both privacy and consumers. Independent wearables like Friend have also raised alarm among reviewers about always-listening behavior without clear opt-out tools. Even though these devices might not necessarily have cameras (yet) to record video footage, they usually come with always-on microphones and can, at the very least, scan for nearby Bluetooth and WiFi devices to collect valuable insights on the user’s surroundings and, more precisely, other users in close proximity. A device category that banks primarily on its video recording capabilities are smart glasses. Unlike the glassholes from a decade ago, this time it seems fashionable and socially accepted to wear the latest cloud-connected glasses. Faced with the very same issues mentioned previously for different device types, smart glasses, too, create immense risks for privacy, with little to no policy in place to protect bystanders . There are several satellite constellations in orbit that house advanced imaging satellites capable of capturing high-resolution, close-up images of Earth’s surface, sometimes referred to as “spy satellites” . These satellites provide a range of services, from military reconnaissance to commercial imagery. Notable constellations by private companies include GeoEye ’s GeoEye-1 , Maxar ’s WorldView , Airbus ’ Pléiades , Spot Image ’s SPOT , and Planet Labs ’ RapidEye , Dove and SkySat . Surveillance tech frequently arrives with a compelling use case, like detering car theft, finding a missing child, automating a customer queue, or making life easier with audio and visual interactions. But it also tends to become infrastructural and persistent. When private corporations, local governments and individual citizens all accumulate recordings, we end up with a mosaic of surveillance that’s hard to govern because it’s distributed across actors with different incentives. In addition, surveillance technologies rarely affect everyone equally. Studies and analyses show disproportionate impacts on already-targeted communities, with increased policing, mistaken identifications from biased models, and chilling effects on protest, religion or free association. These systems entrench existing power imbalances and are primarily benefitial to the people in charge of watching rather than the majority that’s being watched . Ultimately, surveillance not only makes us more visible, but we’re also more persistently recorded, indexed and analyzable than ever before. Each camera, microphone and AI assistant may be framed as a single, sensible feature. Taken together, however, they form a dense information layer about who we are, where we go and how we behave. The public debate now needs to shift from “Can we build this?” to “Do we really want this?” . For that, we need an informed public that understands the impact of all these individual technologies and what it’s being asked to give up in exchange for the perceived sense of safety these systems offer. Avigilon (Motorola Solutions) Axis Communications Bosch Security Systems Sony Professional Axis Communications Bosch Security Systems Flock Safety Kapsch TrafficCom Motorola Solutions (WatchGuard) PlateSmart Technologies Digital Ally Kustom Signals Motorola Solutions (WatchGuard) Transcend Information Flock Safety Lockheed Martin (Procerus Technologies) Quantum Systems Mercedes-Benz Eufy Security Nest Hello (Google) Ring (Amazon) SkyBell (Honeywell) Bystander privacy (how do you notify people they’re being recorded?) Vendor and lifecycle risk (cloud dependence, subscription models, and what happens to device functionality or stored data if a startup folds) Gentle Monster Gucci (+ Snap) Oakley (+ Meta) Ray-Ban (+ Meta) Spectacles (Snap) BAE Systems General Dynamics (SATCOM) Thales Alenia Space
Books
Media
Security
AI
Music
Culture
Hardware
Programming
Entertainment
Writing
Data
Journalism
Business