Quick summary

The North American guard industry is huge, essential, and strained. In the U.S. alone, there were about 1.27 million security guard and gambling surveillance jobs in 2024, with roughly 162,300 openings a year expected through 2034, mostly because workers leave and need to be replaced rather than because the field is exploding with net growth. Median pay for U.S. security guards was $38,370 in 2024. In Canada, median hourly wages for security guards were about C$21.00 nationally, with regional differences across provinces. (Bureau of Labor Statistics)

At the same time, the job is getting harder. Ontario says guards must act professionally, safely, and lawfully, but they are not police and operate within a regulated, limited legal framework. Canada’s Criminal Code allows arrest by a private person only in narrow situations. California similarly requires background checks and mandatory training, including “power to arrest” instruction, before registration. (Ontario)

ArcadianAI’s internal guard-industry materials describe the market well: a large labor-based industry trapped in a high-cost, low-margin cycle, with alert fatigue, liability exposure, and heavy dependence on passive NVR-based systems. Arcadian’s position is that the next step is not replacing guards, but augmenting them with policy-driven intelligence that reduces low-value noise and helps humans focus on events that matter.

Why this topic matters now

Yesterday’s shooting at CF Fairview Mall in Toronto was a brutal reminder that guarding is not an abstract service category. Toronto Police said a mall security team member confronted a fleeing jewelry-store robbery suspect and was shot during the interaction. The guard was later reported to be in stable condition. (Toronto Police Service)

That incident does not define the entire industry, but it does expose the central tension inside it: guards are often asked to be visible, calm, decisive, and effective in moments that can turn violent in seconds, even when the systems around them are reactive rather than intelligent. That tension is at the heart of the guard industry across North America. (Ontario)

Table of contents

1. What the guard industry actually is

2. Market size, jobs, wages, and turnover

3. The main categories of guards

4. The legal reality: guards are not police

5. The strengths of human guarding

6. The structural weaknesses of the industry

7. Why passive video and NVR workflows create blind spots

8. Where AI security monitoring changes the equation

9. What buyers in North America really want

10. A practical comparison: traditional guarding vs AI-augmented guarding

11. FAQs

12. Conclusion and CTA

1) What the guard industry actually is

The guard industry is often misunderstood as a single service. It is really a collection of operating models sold into different risk environments: office towers, retail stores, malls, apartment communities, logistics yards, hospitals, campuses, data centers, public-facing facilities, and high-risk executive or infrastructure settings. U.S. labor data reflects that breadth: security guards work across retail, healthcare, education, accommodation, food services, and specialized guard firms. (Bureau of Labor Statistics)

ArcadianAI’s internal industry analysis breaks the sector into familiar categories: static guarding, mobile patrol, loss prevention, executive protection, and critical infrastructure security. Each category involves different expectations, different liabilities, and different workflow needs.

That matters strategically. A condo concierge, a mall loss-prevention officer, and a critical-infrastructure guard all wear the same broad “security” label, but the economics, training burden, legal risk, and technology requirements are very different.

2) Market size, jobs, wages, and turnover

In the U.S., the Bureau of Labor Statistics says security guards and gambling surveillance officers accounted for 1,272,400 jobs in 2024, with little or no net employment change projected through 2034. But the deeper signal is the expected 162,300 openings each year, which BLS says will mostly come from replacement needs as workers transfer, retire, or exit the workforce. That is one of the clearest public signs that the industry has a retention and churn problem even when top-line employment is stable. (Bureau of Labor Statistics)

Pay levels reinforce the pressure. U.S. median annual pay for security guards was $38,370 in 2024, with median pay in the investigation, guard, and armored car services segment at $37,530. Retail trade security roles were lower, at $36,080. (Bureau of Labor Statistics)

Canada shows a similar labor reality. Job Bank’s latest wage data lists median hourly pay for security guards and related occupations at C$21.00 nationally, with regional variation by province and community. Job Bank’s outlook pages also note that technology is making the occupation “more complex and specialized” even as protection needs continue to support demand. (Job Bank)

The turnover problem is not just anecdotal. A 2025 UC Berkeley Labor Center factsheet found that turnover in New York City’s investigation and security services sector reached 77.0% in 2024, up from 69.3% in 2019, and higher than the private sector overall in the city. The same report found that most guards worked full time, yet median hourly pay was only $20.29, and 80.1% earned less than the MIT living wage benchmark used in the study. (UC Berkeley Labor Center)

ASIS Security Management also reported in 2025 that high turnover was the top challenge named by security service providers in a recent benchmark study, ahead of issues like margins, compliance, and insurance cost. (ASIS International)

What that means

This is not an industry collapsing from lack of need. It is an industry under stress because demand remains strong while the labor model is hard to sustain at scale.

3) The main categories of guards

Static guards

Static guards are the classic on-site officers. They handle access control, patrol rounds, reception-area security, incident documentation, and visible deterrence. They are common in offices, industrial properties, schools, hospitals, and residential buildings.

Mobile patrol

Mobile patrol officers cover multiple sites, often by vehicle. They do lockups, unlocks, alarm response, timed site visits, and exterior checks. This model is cheaper than permanent on-site guarding, but it creates gaps between visits.

Loss prevention and retail guards

These guards focus on theft detection, suspicious behavior, evidence, and store policy enforcement. Retail is one of the clearest examples of the “power gap”: guards are expected to deter and observe, but the legal and liability line around physical intervention can be very narrow.

Concierge, doorman, and residential officers

These roles are part hospitality, part access control, part package and visitor management, and part incident response. In multifamily and condo settings, these personnel often become the visible face of “security,” even though much of their day is operational rather than tactical.

Executive protection

Executive protection is a very different discipline from regular guarding. It combines threat assessment, movement planning, close protection, and more specialized risk management.

Critical infrastructure and high-security site guarding

Ports, utilities, data centers, energy sites, and certain industrial facilities need guards operating in more formal, compliance-heavy, high-consequence environments.

Remote monitoring and SOC-linked guarding

This category matters more every year. Some organizations now blend on-site staff, remote operators, alarm-center workflows, and AI-supported review. This is where ArcadianAI’s positioning is most direct: reducing low-value noise, improving queue quality, and supporting operators with better signal.

4) The legal reality: guards are not police

This is one of the most important truths in the entire industry.

Ontario’s rules say guards are responsible for guarding or patrolling to protect people or property and must act professionally, safely, and lawfully. They must comply with the Private Security and Investigative Services Act and its Code of Conduct. Ontario also restricts equipment use and makes clear that guards may carry certain tools only within specific legal conditions. (Ontario)

In Canada, section 494 of the Criminal Code allows arrest without warrant by a private person only in narrow situations, including when someone is found committing a criminal offence on or in relation to property and the arrest is made immediately or within a reasonable time when police cannot feasibly make it. (Department of Justice Canada)

In California, guards must be at least 18, undergo FBI and DOJ background checks, and complete required training, including “power to arrest” training. They then must complete more training after employment begins and annual training thereafter. (bsis.ca.gov)

ArcadianAI’s internal industry memo summarizes the practical reality well: security guards can remove trespassers under owner authority, use reasonable force for defense where lawful, and in some contexts detain or arrest under narrow rules, but they are not general law-enforcement officers. Their real value is often as deterrents, observers, witnesses, and first responders to the first few minutes of an incident.

5) The strengths of human guarding

For all the pressure on the industry, human guarding still exists for a reason.

Immediate physical presence

A camera can record. A software alert can notify. But a human can walk the site, escort someone out, check a door, lock a gate, help a resident, guide an evacuation, or render immediate aid in a medical emergency. ArcadianAI’s internal analysis correctly lists immediate physical presence as a core strength.

Judgment and de-escalation

Good guards do more than watch. They interpret body language, calm tense situations, recognize when a disturbance is turning serious, and make small decisions that prevent bigger problems. In public-facing sites like malls, apartment towers, and hospitals, this human judgment is often the most valuable part of the service.

Visible deterrence

Many clients buy guards partly because visibility matters. A marked patrol unit, a lobby officer, or a uniformed presence changes behavior. This is hard to quantify but easy to understand.

Incident continuity

When something does happen, guards are often the bridge between the first sign of trouble and law enforcement, EMS, property management, or corporate response.

6) The structural weaknesses of the industry

The industry’s weaknesses are just as real as its strengths.

High turnover

High turnover destroys consistency. New officers do not know the site, do not know the tenants or staff, and do not know the unwritten operating context. It also raises training cost and lowers service continuity. Public labor data and sector reporting both support this as a core challenge. (Bureau of Labor Statistics)

Low margins

ArcadianAI’s internal proposal describes the industry as stuck in a high-cost, low-margin cycle, especially when providers are forced to sell hours rather than outcomes.

Alert fatigue and passive watching

ArcadianAI’s internal materials repeatedly point to the same operational problem: systems detect everything, escalate everything, and overwhelm operators with noise. Monitoring teams begin ignoring low-value activity, and real events can get lost in the queue.

Liability exposure

The more guards are asked to intervene physically, the more risk exists around excessive force, false imprisonment, escalation, and documentation failures. This is especially acute in retail, multifamily, and public-facing settings.

Inconsistent service quality

Guarding quality varies sharply by site, shift, supervisor, and employer. Even the best firms struggle when staffing is thin and turnover is high.

7) Why passive video and NVR workflows create blind spots

A major theme in ArcadianAI’s internal proposal is that many guarding environments still run on passive infrastructure. Cameras record. NVRs store footage. But the system does not really help until after something has already happened.

That problem is bigger than it sounds.

A passive system can be excellent at forensic review and still fail at prevention. In practical terms, that means:

• a break-in is captured but not interrupted

• a trespass is recorded but not escalated in time

• a roaming suspect appears on multiple cameras but is not prioritized

• an operator faces a flood of motion-based or rule-based alerts and misses the one that matters

ArcadianAI’s internal industry note describes this as guards being “blind” until an incident has already occurred.

This is exactly where the guard industry is most vulnerable: the job is increasingly judged in real time, but much of the tooling is still forensic.

8) Where AI security monitoring changes the equation

The strongest version of the AI story is not “replace guards.” It is “stop wasting guard attention.”

ArcadianAI’s internal positioning is that Ranger acts as a policy-driven intelligence layer on top of existing infrastructure, using site-specific rules, schedules, and priorities to surface meaningful events rather than raw noise. It is also described as camera-agnostic, built for recording, review, and multi-site environments, and designed to reduce low-value noise in monitoring workflows.

That matters because guard operations do not need more raw detection. They need:

• better triage

• better context

• better prioritization

• lower false alarm burden

• stronger audit trails

• support for distributed operations

ArcadianAI’s own residential proof point is instructive: over four weeks, across 28 after-hours residential cameras, the system logged 20,210 raw triggers but reduced those to 43 operator-worthy events, filtering out 20,167 low-value events. The key claim is not magic. It is operational signal quality.

The same internal materials describe a 65%–95% reduction in false alarms for existing SOC operations and position the economic benefit as margin expansion, staff retention, and stronger liability posture through audit trails.

Why this matters strategically

Human guards are still necessary in many environments. But human attention is expensive, inconsistent under overload, and impossible to scale well when low-value activity dominates the queue. AI security monitoring is valuable when it helps teams move from static alerts to policy-driven awareness.

9) What buyers in North America really want

Most buyers do not wake up asking for “guards.” They ask for outcomes:

• fewer incidents

• faster response

• visible deterrence

• reduced liability

• better incident records

• more confidence from tenants, staff, shoppers, or executives

• less operational chaos

Guarding companies want better margins and retention. Property owners want more accountability. SOC teams want cleaner queues. Residential operators want after-hours awareness without flooding staff. Retailers want theft visibility without creating more liability. Those are different words for the same thing: more effective human response with less wasted motion.

10) Traditional guarding vs AI-augmented guarding

This is essentially the ArcadianAI thesis in operational form: not fewer humans everywhere, but better use of human time where noise, complexity, and operational pressure are highest.

FAQs

Are guards in North America allowed to intervene?

Yes, but within narrow legal limits that vary by jurisdiction. Guards are generally allowed to protect property, control access, enforce site rules, observe and report, and in some cases detain or arrest under very specific circumstances. They are not general law-enforcement officers. (Ontario)

Is the guard industry growing?

Demand remains strong, but U.S. employment growth is projected to be flat overall from 2024 to 2034. The large number of annual openings reflects replacement needs more than expansion. (Bureau of Labor Statistics)

What is the biggest challenge in the guard industry?

Turnover is one of the clearest structural problems, along with wage pressure, labor shortages, compliance burden, and alert fatigue. (ASIS International)

What kinds of guards are most common?

Static guards, mobile patrol, retail/loss prevention, concierge/residential, executive protection, and critical-infrastructure guards are among the most common categories.

Why are outdated NVR systems a problem?

Because they are mostly passive. They record what happened, but often do not provide meaningful real-time prioritization. That creates blind spots, delayed response, and operator overload.

What is AI security in this context?

In practical terms, AI security means using software to interpret video and events more intelligently so teams can focus on what matters. ArcadianAI’s framing is policy-driven intelligence layered on top of existing video infrastructure.

Quick glossary

AI security: Using artificial intelligence to help detect, prioritize, verify, or interpret security events more effectively.
AI security monitoring: Monitoring workflows that use AI to reduce noise and highlight likely incidents that need human attention.
Cloud NVR / NVR cloud: Video storage and access models that move beyond purely local, hardware-bound recorder workflows.
Loss prevention: Security work focused on reducing theft, shrink, and retail-related losses.
Mobile patrol: Security coverage delivered through scheduled or responsive vehicle-based site visits.
Policy-driven video intelligence: Video analysis that uses site rules, schedules, and priorities rather than simple motion or static alerts alone.
SOC / RVM: Security Operations Center / Remote Video Monitoring, where multiple sites are monitored centrally.

Conclusion: the guard industry is not obsolete, but the old model is under pressure

Security guards are still indispensable across North America. They bring presence, judgment, deterrence, and human response that cameras alone cannot replace. But the business model around them is strained by low wages, high turnover, legal limits, liability pressure, and old surveillance workflows that record everything but help too little in the moment. (Bureau of Labor Statistics)

The future of guarding is not a world with no humans. It is a world where humans are no longer asked to do high-consequence work with low-intelligence systems. That is why the most important shift in the industry is not simply adding more cameras. It is improving signal quality, reducing low-value noise, and giving security teams better context before the situation turns dangerous. ArcadianAI’s own internal materials are consistent on this point: the opportunity is to move from passive, reactive infrastructure to policy-driven awareness that makes guarding more effective, more scalable, and more defensible.

CTA:
If your organization still depends on passive NVR workflows and overloaded human monitoring, now is the time to rethink how security is delivered. ArcadianAI helps teams add a camera-agnostic, policy-driven intelligence layer on top of existing infrastructure so guards and operators can focus on real threats instead of raw noise. Schedule a demo to see how AI security monitoring can improve signal quality, response, and operational visibility.