Menu
Security Guard
Three components - Automation Resistance, Structural Moat, and Demand - add up to the 57.
Physical presence, patrol, access checks, and de-escalation are not text tasks, and measured AI exposure is effectively zero. The watching layer is the exposed part: analytics monitor feeds so fewer posts cover more cameras. Response work holds; watching-only posts consolidate.
Observed AI exposure is 0.00% and the Tufts median job-loss estimate is 1.13% — both at the floor, because the job's core is bodily: patrolling, checking access, removing trespassers, responding to alarms, standing visible at a door. The score sits below the embodied maximum because the monitoring slice of the work — watching feeds, logging, first-draft incident reports — is genuinely software-reachable, and posts built mostly on watching are consolidating.
Camera analytics, anomaly alerts, license-plate readers, and report-drafting tools extend what one guard covers — a single responder plus a monitoring center replaces several static posts. The staffing savings flow to the contract and the client, not the guard's rate; the worker upside is modest and concentrated in operations-center and supervisor roles that run the technology.
Most states license the work and liability keeps a person on site, but the licensing bar is a course, not a credential. Patrol robots handle routes, not judgment. Armed endorsements and state training-hour rules are where the bar actually rises.
Federal physical data shows the job is presence more than exertion: standing about 4.7 hours a shift, light mean lifting around 8 pounds, with substantial seated monitoring time. The physical reality that matters is location and risk — outdoor rounds, night posts, and confrontation exposure — which keeps a human on site but adds only moderate protection by itself.
Most states require a guard license or registration — federal data shows about 58% of the workforce holding a required license, certification, or registration, far above typical service work. The moat is real but shallow: a course measured in hours-to-days plus a background check, with armed endorsements adding a genuine second gate where states allow them. It restricts who may hold the post without making the skills scarce.
Patrol robots and camera drones are deployed at the margins — fixed routes in garages, campuses, and warehouses — and they extend coverage rather than replace response. A machine can flag a propped door; removing the person propping it, de-escalating them, and writing the report a court will accept remains human work. Unstructured environments, weather, and confrontation keep replacement resistance near the top.
High school plus a state licensing course is the standard entry, and O*NET places the occupation in Job Zone 2. Depth accrues through add-ons — armed endorsement, medical certifications, access-control systems — but the base credential is shallow by design, which is what keeps entry fast and wages near the service floor.
Need persists in every economy and liability keeps a staffing floor, but remote monitoring thins posts per site and turnover supplies most openings. Insurance requirements hold the floor; churn in contract patrol sits on top of it.
Federal projections count about 1.26 million jobs and about 161,000 annual openings with essentially flat projected employment (+0.4%). The occupation is large and hires constantly; the opening rate reflects high turnover in contract work more than expansion.
Openings are dominated by replacement churn on the contract side, where pay floors and rotating sites push constant exits. The structural buyers — hospitals, campuses, utilities, corporate facilities — hire steadily but in smaller numbers. Demand built mostly on turnover scores low even when the underlying need is permanent.
Security spending holds through downturns — properties, liability, and insurance requirements do not pause — and crime concerns tend to raise demand when other hiring falls. The offset is the monitoring consolidation already underway: each analytics generation covers more feeds per human, so flat employment against a growing building stock is the realistic path rather than growth.
The case weakens if analytics plus monitoring centers become the default for ordinary buildings, leaving on-site guards only at high-risk properties. The threshold is insurer and client behavior: watch whether liability standards start accepting remote-plus-response coverage in place of a posted guard.
The case improves if more states expand required training hours, armed standards, and continuing education — the pattern several large states have followed. Higher bars raise the wage floor, cut churn, and push the occupation toward the professional in-house tier.
The career case improves if the occupation keeps tilting from static watching toward trained response — medical, de-escalation, incident command — integrated with monitoring technology. Fewer posts, better jobs: the remaining guards earn more and carry transferable public-safety skills. State training expansions would accelerate the shift.