Surveillance was once imagined as a camera watched by a police officer in a control room. Today, it is an interconnected commercial ecosystem combining cameras, licence-plate readers, facial recognition, artificial intelligence, cloud storage, digital evidence platforms and enormous searchable databases.
Private companies increasingly build and operate the infrastructure through which governments, police agencies and businesses observe the public. The question is no longer simply who is watching—it is who owns the technology, controls the information and profits from its expansion.
Report by eLocal
The Watchers Are Increasingly Private Companies
In Part 1 of this series, Who Watches The Watchers?, eLocal examined the democratic and constitutional challenge created when surveillance capabilities expand faster than public oversight. Part 2 follows the infrastructure itself.
The modern surveillance system is no longer built exclusively inside government departments. Much of it is designed, supplied, maintained and stored by commercial technology companies. Police and public agencies may operate the systems, but private vendors frequently provide the cameras, software, artificial intelligence, cloud storage, analytics and technical support.
This has created an industry whose commercial success can depend upon the continuing expansion of surveillance.
Companies are not necessarily doing anything improper by selling security technology. Police need reliable evidence systems. Retailers need ways to address repeat offending. Emergency services need accurate information, and investigators need tools capable of handling large volumes of digital evidence.
However, the commercial incentives remain important.
- A company selling cameras benefits when more cameras are installed.
- A cloud provider benefits when more data is collected and stored.
- A software platform benefits when more agencies, businesses and databases become connected to its network.
The public interest may favour targeted, proportionate and demonstrably effective surveillance. The commercial interest may favour broader adoption, longer contracts, more connected devices and deeper dependence upon a vendor’s ecosystem.
Those interests are not always identical.
From Cameras to Networks
The traditional security camera recorded pictures. Modern surveillance systems create searchable information.
Automatic Number Plate Recognition (ANPR) uses optical character recognition to identify vehicle registration plates. New Zealand Police says it uses ANPR for enforcement, staff safety and public-safety purposes.
A modern plate-reading system may record more than the plate itself. Depending upon the product, it can associate the plate with a location, date, time, vehicle description and direction of travel. When thousands of cameras contribute to the same network, investigators can potentially reconstruct where a vehicle has travelled and identify vehicles appearing near multiple locations.
Facial-recognition systems operate differently. They analyse facial characteristics and compare them with images held in a database or watchlist. New Zealand’s Privacy Commissioner describes biometric identification as an automated process that asks whether a person is known or appears in a database.
Other systems combine video feeds, police records, emergency calls, vehicle information, retail reports, maps and live operational data within a single interface. Artificial intelligence can be used to sort footage, identify patterns, generate alerts or assist with report writing.
The transformation is fundamental.
A camera no longer needs to be watched continuously by a human. The system can watch, classify and search on their behalf.
Axon: From Body Cameras to an Operating System for Policing
Axon Enterprise is widely associated with TASER devices and police body-worn cameras. Its business now extends far beyond either product.
Axon describes an integrated suite incorporating body cameras, in-vehicle cameras, digital evidence management, real-time operations software, drones, robotic security, training tools and cloud-based services.
Its Axon Evidence platform allows law-enforcement agencies to store, manage, review and share digital evidence through a cloud-based system. Body-camera footage can therefore become part of a much larger evidence environment containing photographs, interview recordings, vehicle footage, documents and other investigative material.
Axon has also developed Draft One, which uses artificial intelligence and transcripts of body-camera audio to produce an initial draft of a police report.
The potential efficiency benefit is obvious. Police officers frequently spend considerable time completing administrative work. A system that produces a usable first draft could release officers for frontline duties.
But it also raises significant questions:
- How accurately does artificial intelligence interpret unclear speech, stressful encounters or competing versions of events?
- Will officers carefully verify every generated statement?
- Could the language selected by an algorithm subtly influence the official account of an incident?
The issue is not simply whether an AI-generated report contains a blatant error. It is whether automated drafting may shape what is emphasised, omitted or normalised within the evidential record.
Axon’s financial growth shows the value of integrating devices with recurring software subscriptions. The company has reported rapidly growing annual recurring software revenue built around subscriptions, premium products and expanding digital services.
This illustrates the changing economics of surveillance.
- The camera may be purchased once.
- The software, storage, licences and integrated services can generate continuing revenue for years.
Flock Safety: Building a Searchable Map of Vehicle Movements
Flock Safety has grown through automated licence-plate readers and connected public-safety camera networks.
The company says its cameras collect vehicle information rather than biometric information and that searches are restricted to approved users, with access logged for review. It also states that licence-plate information is normally retained for thirty days before deletion, unless different legal requirements apply.
These safeguards matter.
Yet retention is only one part of the privacy question. Even a database kept for thirty days can contain an extensive record of ordinary people:
- Travelling to work.
- Visiting relatives.
- Attending political meetings.
- Receiving medical treatment.
- Participating in lawful protests.
The information may be individually unremarkable. Its power emerges through aggregation.
- One camera records a plate at one location.
- A network of cameras may show a pattern of movement.
The distinction between investigating a vehicle linked to an identified crime and searching the movements of thousands of innocent motorists is therefore crucial. The technology can be the same; the purpose, threshold and oversight may be entirely different.
The constitutional issue is not whether licence-plate readers can help solve serious crimes. They clearly can.
The issue is what rules prevent a targeted investigative tool from becoming a general system for recording population movement.
Palantir: Connecting Databases That Were Once Separate
Palantir Technologies represents another part of the surveillance ecosystem: the integration and analysis of data.
Its Gotham platform is designed to combine information from multiple government and operational databases, allowing users to visualise relationships, identify patterns and support operational decision-making.
Those capabilities can be extremely valuable when investigating organised crime, terrorism or large-scale fraud.
However, the greatest surveillance power may not come from any individual database.
It may arise when separate datasets are connected:
- A vehicle record may reveal movement.
- A police report may contain allegations.
- A retail platform may record suspected offending.
- A government database may confirm identity.
- A facial image may provide a potential match.
- A telecommunications record may establish contact between individuals.
When those datasets are isolated, their usefulness and their risk are limited. When they are combined, they can create a remarkably detailed picture of a person’s relationships, movements, behaviour and daily life.
The ability to connect information can expose organised crime and improve investigations.
It can also magnify errors.
The more convincing the technology appears, the more important it becomes that users understand its limitations.
Auror: New Zealand’s Retail-Crime Intelligence Network
The surveillance business is not solely foreign.
New Zealand-founded Auror has developed a retail-crime intelligence platform now used by retailers and law-enforcement organisations internationally.
Auror allows retailers to record incidents, identify repeat offending patterns, package evidence and share selected information with police. It also incorporates analytics, vehicle recognition and AI-assisted matching technologies supported by human verification.
The potential benefit is clear.
A person committing repeated offences across several locations may be difficult for any individual retailer to identify. A shared platform can reveal patterns and provide investigators with consolidated evidence.
The danger is that private suspicion may begin to operate like a quasi-official intelligence record.
Retail employees are not police investigators. An incident report may contain incomplete information, uncertain identification or subjective interpretation. A person may be labelled as suspicious without ever being charged or convicted.
That does not make shared retail intelligence inherently unacceptable.
It means standards of accuracy, correction, access, purpose and retention become critical.
Citizens deserve to know:
- What information is held about them.
- How long it is retained.
- Who can access it.
- How inaccuracies can be corrected.
Following the Data
Every surveillance system creates a chain of custody.
- Who collects the information?
- Who owns it?
- Where is it stored?
- Who can search it?
- Who can share it?
- How long is it retained?
- Can it be copied into another system?
- Can an individual discover that a record exists and challenge it?
These questions are often less visible than the cameras themselves. Yet they may be more important.
The public is regularly assured that access is restricted and logged.
Those safeguards are valuable.
But an audit trail only protects the public when:
- Audits are actually conducted.
- Improper searches are detected.
- Meaningful consequences follow.
The existence of a technical safeguard should never be confused with proof that governance is effective.
New Zealand’s Expanding Framework
New Zealand Police has publicly released information about its use of emerging technologies, including ANPR, facial recognition and the Auror platform. It has also established an external expert panel to provide ethical and policy advice on future technology deployments.
The Office of the Privacy Commissioner has introduced New Zealand’s Biometric Processing Privacy Code, establishing specific privacy rules governing the collection and use of biometric information.
These developments demonstrate that governance is beginning to catch up with technology.
But policies, privacy codes and advisory panels do not remove the need for democratic scrutiny.
Surveillance affects the relationship between the citizen and the state.
That makes it a democratic issue.
Does It Work?
The most important question may also be the least consistently answered.
Does surveillance produce measurable improvements in public safety?
It is not enough to identify individual cases where a camera or database assisted an investigation. Any information-gathering system will occasionally prove useful.
The proper evaluation must compare the benefit with the overall cost:
- How many serious offences were prevented?
- How many investigations were successfully concluded?
- Did crime actually decrease?
- Were missing people located more quickly?
- Did child-exploitation investigations improve?
- Were response times reduced?
- How many innocent people were affected?
- Were less intrusive alternatives available?
These are not anti-police questions.
They are the questions every accountable public institution should expect to answer before expanding powerful surveillance capabilities.
The Problem of Function Creep
Surveillance systems are frequently introduced for a narrow and apparently compelling purpose:
- A camera network may begin with stolen vehicles.
- Facial recognition may initially target serious violent offenders.
- Retail intelligence may focus on organised retail crime.
Once the infrastructure exists, however, pressure often grows to extend its use.
This is known as function creep.
A system introduced for one purpose gradually acquires new purposes.
The safeguard is not a promise that misuse will never occur.
The safeguard is a legal framework requiring any significant expansion to be:
- Publicly justified.
- Independently reviewed.
- Democratically authorised.
When Public Infrastructure Becomes Vendor Infrastructure
Another issue is dependence.
When agencies adopt integrated surveillance platforms, evidence storage, cloud infrastructure and operational software may all become linked to the same commercial supplier.
Changing vendors later can become technically difficult and financially expensive.
Procurement decisions made today may shape policing infrastructure for decades.
New Zealand therefore has a legitimate interest in understanding whether strategically important policing systems remain under practical national control.
From Smart Cities to 15-Minute Cities
The surveillance infrastructure examined throughout this report also intersects with a broader international movement towards so-called “smart cities” and the increasingly discussed concept of the 15-minute city.
The 15-minute city is an urban-planning model in which residents can access most of their daily needs—including work, education, healthcare, shopping and recreation—within approximately fifteen minutes by walking, cycling or public transport.
Supporters argue that such communities can:
- Reduce congestion.
- Improve public health.
- Lower emissions.
- Strengthen local neighbourhoods.
- Reduce dependence on private vehicles.
The concept itself is not expressly identified as a pillar of the United Nations’ original 2030 Agenda for Sustainable Development. However, it aligns closely with Sustainable Development Goal 11, which seeks to make cities and human settlements “inclusive, safe, resilient and sustainable.”
Over recent years, the concept has also been promoted and discussed by organisations including UN-Habitat, the World Economic Forum, and various international urban-planning networks as one possible approach to achieving those broader sustainability objectives.
None of this means that walkable communities are inherently instruments of surveillance or social control. Throughout history, towns and villages have naturally developed around local schools, shops, parks and community services. Few New Zealanders would object to having essential services located closer to where they live.
The constitutional question arises not from the urban-planning concept itself, but from the digital infrastructure increasingly proposed alongside modern smart-city developments.
Today’s smart-city technologies can include:
- Interconnected CCTV networks.
- Automatic Number Plate Recognition.
- Facial-recognition systems.
- Traffic sensors.
- Environmental monitoring.
- Mobile-phone location data.
- Digital identity platforms.
- Smart transport systems.
- Cloud computing.
- Artificial intelligence capable of analysing enormous volumes of information in real time.
Viewed individually, many of these technologies serve legitimate and useful purposes:
- Traffic cameras may improve road safety.
- Environmental sensors can monitor air quality.
- Smart transport systems may reduce congestion.
- Police technology can assist investigations and help locate dangerous offenders.
The question is what happens when these systems become integrated.
As individual technologies become connected through shared databases and artificial intelligence, they have the potential to build increasingly detailed pictures of how communities function and how individuals move through public spaces.
Used transparently, proportionately and within clearly defined legal limits, these technologies may deliver genuine public benefits. Combined without sufficient safeguards, however, they could also create the technical capability to:
- Monitor movement.
- Identify individuals.
- Analyse behaviour.
- Automatically enforce rules across entire communities.
The important distinction is therefore between proximity and control.
Designing neighbourhoods where people can conveniently reach essential services is fundamentally different from creating digital systems capable of continuously recording:
- Where people travel.
- Who they associate with.
- Whether access to services depends upon digital credentials.
New Zealand has an opportunity to learn from international developments before making long-term decisions about its own urban infrastructure.
If smart-city technologies are introduced, they should be assessed not only for their environmental and operational benefits, but also for their impact on:
- Privacy.
- Freedom of movement.
- Democratic accountability.
- The potential for future function creep.
A system introduced today to improve traffic management may tomorrow be capable of supporting entirely different functions.
History shows that technology rarely stands still:
- Capabilities expand.
- Software evolves.
- Data accumulates.
The legal framework must evolve just as quickly.
Ultimately, New Zealanders deserve an open public conversation about where the line should be drawn.
- Technology should assist communities.
- Communities should never exist simply to feed technology.
Applying the National Interest Test
eLocal’s National Interest Test asks one simple question:
Does this make New Zealand stronger?
Surveillance technologies can strengthen national resilience when they help:
- Protect frontline staff.
- Locate dangerous offenders.
- Dismantle organised criminal networks.
- Provide reliable evidence before the courts.
They can weaken democratic resilience when they:
- Enable unaccountable data collection.
- Normalise population tracking.
- Concentrate sensitive information in private commercial systems.
- Expand beyond their original purpose without meaningful public debate.
- Operate without measurable evidence of public benefit.
The appropriate response is neither blind acceptance nor automatic rejection.
Every significant surveillance system should be required to demonstrate:
- A clear and lawful purpose.
- A defined threshold for use.
- Evidence that the system is effective.
- Independent privacy and human-rights assessments.
- Strict data-access and retention rules.
- A process for correcting inaccurate information.
- Regular audits with publicly reported findings.
- Transparent procurement and total lifecycle costs.
- Human review of automated decisions.
- Clear penalties for misuse.
- A sunset or formal review mechanism.
New Zealanders should not have to prove that a surveillance system is dangerous before safeguards are introduced.
The institution proposing the system should demonstrate that it is necessary, proportionate, transparent and effective.
Only then can the public have confidence that technology remains accountable to democracy rather than democracy gradually becoming accountable to technology.
Who Owns the Watchers?
The rise of private surveillance companies changes the original question.
Who watches the watchers when the watchers are no longer a single institution, but an interconnected network of:
- Police agencies.
- Retailers.
- Councils.
- Technology vendors.
- Cloud providers.
- Artificial-intelligence systems.
Responsibility can become dispersed until every participant controls one part of the system while no participant accepts responsibility for the whole.
Oversight must therefore follow the data itself.
Citizens deserve to understand:
- How the network operates.
- Who has access.
- What measurable public benefit it delivers.
Final Thought
The companies building today’s surveillance infrastructure do not need to conspire.
They simply build products that governments, businesses and institutions increasingly wish to purchase.
The system grows:
- Contract by contract.
- Camera by camera.
- Database by database.
- Subscription by subscription.
Each camera may appear reasonable. Each software upgrade may promise efficiency. Each database connection may solve a legitimate operational problem.
Yet the cumulative result can gradually create a society in which movement, identity, association and behaviour become increasingly observable through interconnected digital systems.
The future of surveillance will not be determined by one dramatic vote.
The question New Zealand must answer is not whether technology should improve public safety.
It is whether public-safety technology will remain the servant of a free society—or quietly become the infrastructure through which that society is permanently observed.
Coming Next
Part 3 — Digital Identity: Convenience or Control?
The next instalment of eLocal Investigates: Surveillance State will examine New Zealand’s emerging digital identity infrastructure, online age verification, digital credentials and the wider constitutional questions surrounding identity, privacy and democratic consent.
Sources
This report draws upon publicly available material published by New Zealand Police, the Office of the Privacy Commissioner, Axon Enterprise, Flock Safety, Palantir Technologies, Auror, United States Securities and Exchange Commission filings and other primary-source documentation referenced throughout the investigation.
Independent reporting. Original context. Credited sources.