The Surveillance of Homeless Populations in Smart Cities
Smart city projects are often presented as practical responses to urban problems. Networked cameras, environmental sensors, connected lighting, predictive software, and public Wi-Fi promise safer streets and better services. For people with stable housing, these systems can seem like background infrastructure. For people living in shelters, vehicles, encampments, or on the street, they can become a constant presence in daily life.
The surveillance of homeless populations by smart city initiatives deserves particular scrutiny because homelessness already involves intense institutional observation. People may be required to identify themselves to access accommodation, food, healthcare, transport, or benefits. When digital systems are added to these points of contact, a person’s movements, relationships, habits, and eligibility can become easier to record and harder to escape.
This is not an argument against every technological tool used by municipal services. Accurate information can help allocate beds, coordinate outreach, and identify urgent medical needs. The concern is that systems designed for efficiency can quietly become systems of control, especially when people have little power to refuse data collection or challenge decisions made about them.
When Public Space Becomes A Data Source
A smart city treats the urban environment as something that can be measured continuously. Cameras count pedestrians, license-plate readers track vehicles, mobile networks generate location records, and sensors monitor noise, air quality, traffic, and footfall. Some projects combine these sources with police records, social-service databases, and privately purchased information.
Homeless people are especially exposed because they often spend much of their lives in public or semi-public places. A person sleeping under a bridge, charging a phone in a library, or moving between transit stations can be visible to several systems at once. A camera may register a body, a Wi-Fi network may log a device, and a municipal worker may add a note to a case-management file. These fragments can create a detailed profile even when no single system appears especially invasive.
Visibility also shapes how a city defines a problem. A person resting on a pavement may be interpreted as an obstruction, a safety risk, or an indicator of disorder. The technology does not make that judgment by itself; policies and human operators determine what counts as an unacceptable presence. Automated counting can give these choices an appearance of neutrality while directing enforcement toward people who have the fewest private spaces available.
From Service Coordination To Social Control
Data sharing between housing departments, police, hospitals, transit authorities, and nonprofit providers is commonly justified as coordination. If organizations know that someone has already received an assessment or emergency treatment, they may avoid repeating paperwork. A shared record can help prevent people from being sent from office to office while in crisis.
The same infrastructure can produce harmful secondary uses. Information collected to offer shelter may later support anti-camping enforcement, warrant checks, immigration scrutiny, or restrictions on access to public facilities. A database built for care can become an instrument for monitoring compliance. Data retention makes this shift especially difficult to detect because records may remain available long after the original service interaction.
Predictive analytics introduces another layer of risk. A model might identify locations where outreach workers should be deployed, but it may also label neighborhoods as sources of disorder or assign individuals a higher risk score based on prior contacts. Historical records are rarely neutral. If some communities have received more police attention or fewer housing services, an algorithm can treat that unequal history as evidence of future risk.
The possibility of error is serious. A duplicated identity, outdated address, mistaken incident report, or misclassified encounter can affect a person’s access to shelter and benefits. People without stable addresses often lack the documentation, time, and legal support needed to correct a record. The result is a form of administrative invisibility: a person is highly visible to surveillance systems but difficult to recognize as an authority over their own data.
Consent Is Weak When Choices Are Limited
Consent has a different meaning when a person must accept monitoring to obtain essential support. A shelter may require identification, a service provider may use biometric check-in, or a public charging station may demand a phone number. Technically, someone can decline. Practically, declining may mean losing access to warmth, electricity, communication, or a bed.
This imbalance complicates the language of voluntary participation. A lengthy privacy notice cannot repair a situation in which the alternatives are unsafe or unavailable. Nor does consent to one purpose automatically justify every later use. Someone who provides information to receive a housing referral has not necessarily agreed to location tracking, facial recognition, or predictive policing.
Digital exclusion adds pressure. Many public services now depend on online accounts, text-message verification, app notifications, and automated scheduling. People whose phones are lost, shared, disconnected, or frequently replaced may have trouble maintaining access. A system that records users more aggressively can still serve them poorly, while creating a permanent trail of missed appointments and failed logins.
A rights-based approach therefore treats privacy as part of service quality. People need clear explanations, meaningful alternatives, access to their records, and a way to refuse unnecessary collection without being punished. Privacy should not be framed as a luxury reserved for residents with homes, bank accounts, and reliable internet connections. The broader question of personal exposure is explored in personal threat model, a useful lens for thinking about risks without assuming that every threat is equally likely.
Comparing Tools And Their Risks
Smart city technologies vary widely in purpose and danger. A sensor that measures air pollution does not carry the same consequences as a facial-recognition system. Yet even apparently harmless tools can become identifying when combined with other records. Risk depends on what is collected, how long it is stored, who can access it, and whether people can challenge its use.
| Technology | Possible public benefit | Risks for unhoused people | Basic safeguard |
|---|---|---|---|
| Environmental sensors | Measures heat, air quality, or noise | Can be paired with location patterns and enforcement data | Collect aggregate data wherever possible |
| CCTV and video analytics | Supports incident investigation and emergency response | Tracks presence, association, and routines in public spaces | Ban facial recognition and limit retention |
| Wi-Fi and Bluetooth tracking | Helps manage transport or facility capacity | Identifies devices and repeated visits without clear consent | Use anonymous, short-lived identifiers |
| Service databases | Coordinates housing, health, and outreach programs | Enables profiling, denial of services, or cross-agency surveillance | Separate purposes and restrict access |
| Predictive analytics | Directs outreach resources | Reproduces biased policing and service histories | Require audits, human review, and appeal rights |
| Biometric check-in | Prevents duplicate registrations or fraud | Creates sensitive identifiers that are difficult to replace | Prefer non-biometric alternatives |
The table illustrates why broad claims about “smart technology” are insufficient. A city can reject facial recognition while retaining extensive camera footage. It can remove names from a dataset while preserving unique movement patterns that make people identifiable. It can publish a privacy policy while allowing contractors to reuse records for commercial development.
Technology assessments should therefore examine the whole data lifecycle. Collection is only the first stage. Storage, linking, sharing, analysis, retention, and deletion each create separate opportunities for misuse. Procurement contracts matter as much as technical specifications because vendors may receive access to operational data or seek to expand a pilot into a permanent system.
Designing Services Around Dignity
The safest principle is data minimization: collect the smallest amount of information needed for a clearly defined service, and delete it when that purpose ends. A city coordinating emergency shelter may need occupancy numbers and basic accessibility requirements. It may not need a detailed history of every location a person has visited.
Purpose limitation should be concrete rather than symbolic. A policy should state whether information can be shared with police, immigration authorities, landlords, debt collectors, researchers, or private technology companies. Exceptional access should require documented authorization, and emergency provisions should be reviewed afterward. Vague language about “public safety” is too broad to protect people whose presence is already treated as suspicious.
Design choices can reduce exposure. Services can use anonymous tokens instead of names for routine visits, offer paper and in-person options, and avoid collecting phone numbers when they are unnecessary. Independent advocates should be present when people are asked to provide sensitive information. People should receive receipts or plain-language summaries showing what was collected and how to request correction or deletion.
Participation must extend beyond consultation after a project has already been selected. People who have experienced homelessness should help define the problem, assess proposed tools, and monitor outcomes. Their expertise includes knowledge of how rules are enforced in practice, which locations feel safe, and how data errors affect access to services. A system that looks efficient in a procurement document may be damaging at street level.
Rules That Put Rights Before Efficiency
Cities need enforceable limits rather than promises of responsible innovation. These limits should cover public agencies, contractors, nonprofits operating under municipal agreements, and vendors that process city data. If a private company performs a public function, outsourcing should not remove public accountability.
Independent oversight is especially important where people cannot easily opt out. Auditors should be able to inspect source data, model performance, access logs, vendor contracts, and deletion practices. Reports should identify how systems affect different groups, including people sleeping outdoors, people with disabilities, migrants, and those experiencing mental-health crises. Oversight without access to meaningful evidence becomes public relations.
People also need remedies. They should be able to know whether a record exists about them, obtain a comprehensible copy, correct inaccuracies, and appeal decisions influenced by automated processing. Complaints must be possible without a permanent address or reliable internet connection. Legal aid, ombuds offices, and community organizations can provide support where individual self-advocacy is unrealistic.
The strongest safeguard may be a presumption against monitoring that does not produce a clear, demonstrable benefit. A city should explain why a tool is necessary, why less intrusive alternatives are inadequate, and how success will be measured. If a pilot cannot show improved housing access, health outcomes, or safety without disproportionate privacy costs, it should end rather than become permanent through inertia.
Practical Standards For Humane Smart Services
Municipal leaders, service providers, and technology teams can translate these principles into operational decisions:
- Prohibit facial recognition, emotion detection, and biometric identification in homelessness-response programs.
- Separate care records from policing systems, with narrow exceptions subject to independent review.
- Publish data maps that show what is collected, by whom, for what purpose, and for how long.
- Provide non-digital and non-tracking ways to access shelter, benefits, charging, transport, and appointments.
- Create paid advisory roles for people with lived experience of homelessness and give them authority over project decisions.
These standards should apply before procurement, not after a system has been installed. A privacy impact assessment should examine likely effects on people who have no private home, no consistent device, and limited ability to contest official records. It should also consider whether a proposed intervention addresses housing insecurity or merely makes it less visible to housed residents.
The distinction matters because surveillance can improve the appearance of order while leaving the underlying crisis untouched. Moving people away from monitored commercial districts, installing hostile architecture, or using analytics to target encampments may change the data without improving anyone’s security. Smart infrastructure becomes socially useful when it supports housing, healthcare, sanitation, and voluntary outreach rather than optimizing exclusion.
Technology policy is part of a larger political choice about who is entitled to remain visible in a city. The broader writing at Twenty of Time examines privacy, digital rights, and the social effects of systems that turn ordinary life into measurable data. Those concerns belong in homelessness policy because public space is where surveillance and inequality meet most directly.
A city that wants to be intelligent should measure success by reduced displacement, stable housing, improved health, and greater autonomy. Residents, outreach workers, civil-liberties groups, and people with lived experience can press officials to publish surveillance inventories, challenge intrusive pilots, and demand services that work without coercive data collection. Public technology should help people regain control over their lives, not make their loss of privacy the price of being seen.