Short answer: concealed weapons detection technology can help a trained screening team identify items that require follow-up, but no portal, sensor, algorithm, or artificial intelligence system detects every prohibited item in every operating condition. The useful comparison is not which product sounds most advanced. It is which complete screening process can produce acceptable results for the facility’s defined threat scope, traffic, staffing, accessibility, privacy, and response requirements.
This guide explains the major technology categories, what an alert means, where AI can assist, which performance claims need evidence, and how to test a system before approving it. Organizations evaluating a New Jersey deployment can also review Rivell’s concealed weapon detection system planning service.
How this guide was prepared
Reviewed: August 18, 2026. The comparison uses primary material from the U.S. Department of Homeland Security, CISA, the National Institute of Justice, NIST, the Federal Trade Commission, the U.S. Department of Justice, and current manufacturer documentation. Government sources define technology boundaries, screening operations, AI risk, accessibility, and substantiation requirements. Manufacturer sources describe current product functions, not independent proof of performance.
- Source policy: operational and legal statements link to the responsible government body; product descriptions link to the manufacturer.
- Evidence rule: a throughput, detection, false-alarm, labor, or comparison claim should be accepted only when the buyer can inspect the test method, configuration, objects, population, environment, exclusions, and results.
- Freshness rule: verify current model, software, subscription, integration, support, and test documentation before procurement because product capabilities and terms change.
- Decision rule: approve the complete people, process, and technology workflow, not the sensor by itself.
What concealed weapons detection means
Concealed weapons detection is an umbrella term for equipment and processes used to find items that a facility defines as weapons, contraband, or prohibited objects. The older but still useful National Institute of Justice technology guide explains that different systems use different physical principles and have application-specific limits. A detection system does not determine intent, establish that an item is legally possessed, or replace an approved search and response policy.
The current DHS SAVER market survey groups commercial walk-through systems around metal detection and millimeter-wave imaging. It also states that the included product information was not independently verified by SAVER. That distinction matters: a market survey can show what vendors offer, while a buyer’s pilot must prove what a proposed configuration does in the intended environment.
Concealed weapons detection technology compared
| Technology | What it evaluates | Useful role | Important boundary |
|---|---|---|---|
| Walk-through metal detector | Changes in an electromagnetic field associated with metallic objects | Primary screening at a controlled checkpoint | Object composition, orientation, sensitivity, nearby metal, electrical conditions, and traffic can affect results. |
| Handheld metal detector | Localized metallic response near the wand | Focused follow-up after an alert or as a defined primary process | Results depend heavily on operator technique, coverage, distance, pace, and the approved follow-up procedure. |
| Millimeter-wave or imaging system | Reflected energy and software-derived information about objects or anomalies | Screening where the defined threat set or operating model requires more than conventional metal response | Buyers must verify object scope, environmental limits, image or alert handling, privacy controls, and secondary screening. |
| Sensor fusion with algorithmic classification | Multiple sensor signals interpreted by software to flag a potential item or location | Higher-flow screening and more targeted operator review when validated for the use case | AI or automation does not eliminate misses, nuisance alarms, sensitivity tradeoffs, staffing, or human confirmation. |
| Bag screening | Contents of bags or carried items through a separate inspection method | A layered process for objects that should not pass through a people-screening lane | A people-screening portal and a bag-screening system are not interchangeable. Route, custody, operator qualification, and prohibited-item procedures still apply. |
DHS’s metal-detector technology note explains the difference between passive and active metal detection and the common relationship between walk-through and handheld screening. It is foundational material, not a current product catalog. Use it to understand the physics, then require current model documentation and site-specific evidence.
What AI changes, and what it does not
AI-enabled systems may classify sensor patterns, distinguish some personal items from some threat patterns, localize an alert, or help an operator review system information. Those functions can be useful, but the word AI is not a performance result. The NIST AI Risk Management Framework calls for governed roles, context mapping, measurement, and ongoing risk management. For a screening application, that means defined human authority, repeatable tests, documented uncertainty, monitoring, change control, and a response when performance falls outside approved limits.
The FTC’s Evolv Technologies action is a direct warning against accepting unsupported claims about detecting all weapons, ignoring harmless personal items, accuracy, false alarms, screening speed, labor cost, testing, or AI performance. Treat the allegations and settlement as substantiation guidance, not as a substitute for evaluating the current product. Ask for the exact evidence behind every proposed performance statement.
Current Avigilon product material describes sensor and software functions, adjustable sensitivity, alert-location information, and video integration. Its current data sheet routes alerts to secondary screening. That is the right operating assumption: a system alert starts a trained review; it is not a final decision.
Detection, nuisance alarms, and throughput are connected
Changing sensitivity can change both what the system flags and how often people or belongings require follow-up. A higher-flow portal does not guarantee a higher-flow entrance if secondary screening, bag diversion, communications, accessibility accommodations, or exception handling becomes the bottleneck. Measure the complete checkpoint from arrival through final disposition.
Before comparing vendors, define the approved test objects and scenarios, how they may be used safely, representative personal items, lane conditions, environmental conditions, population, operator experience, and what counts as a pass, alert, cleared item, miss, nuisance alarm, unresolved event, or process failure. Record results by configuration. Do not combine results from different sensitivity settings or operating models into one headline number.
When a screening system may or may not fit
A walk-through system may be worth evaluating when a facility has a defined prohibited-item policy, controlled entry points, predictable traffic, space for queues and secondary screening, trained staff, and authority to act on the result. It may also support temporary events, but temporary use still needs power, layout, weather, transport, calibration, staffing, communication, and fallback planning.
The equipment may be a poor fit when people can enter through uncontrolled routes, the organization cannot staff follow-up screening, the threat definition is vague, the checkpoint would obstruct emergency egress, privacy or accessibility requirements have not been designed into the process, or leadership expects the portal to replace broader prevention and response controls. A failed fit assessment is useful evidence. It can prevent a costly installation that the operating environment cannot support.
Do not force every location into one configuration. A school morning arrival, hospital entrance, municipal building, workplace, stadium event, and private tenant area can have different traffic, belongings, authority, risk, accessibility, and response conditions. A multi-site organization should document each operating profile and prove any proposed standard against representative sites before repeating it.
Compare proposals using the same operating scenario
Require each provider to respond to the same facility facts and acceptance plan. Separate equipment price from subscriptions, accessories, freight, installation, electrical and network work, site preparation, training, staffing, bag screening, secondary-screening equipment, maintenance, software updates, support, replacement, expansion, data export, and contract exit. A lower device price does not establish a lower operating cost, and a faster portal specification does not establish a faster checkpoint.
Normalize the evidence as well as the cost. Identify whether a claimed result came from a laboratory, manufacturer demonstration, customer environment, commissioned test, independent assessment, or the buyer’s own pilot. Record who selected the objects and settings, who observed the test, and whether misses and nuisance alarms were disclosed. When two proposals use different threat sets or sensitivity settings, their headline figures are not directly comparable.
The screening program matters as much as the equipment
CISA’s Facility Access Control guidance describes screening as an integrated process with trained personnel, initial and follow-up procedures, equipment testing, calibration, and operator interpretation. CISA’s Public Venue Security Screening Guide adds patron communication, prohibited-item handling, staffing, checkpoint design, searches, training, and escalation.
The program should define the prohibited-items policy, authority, entrances, hours, exceptions, accessible route, bag process, refusal process, secondary-screening location, supervisor role, law-enforcement contact, documentation, privacy, emergency egress, bypass prevention, equipment faults, power or network interruption, and return to normal operation. Technology cannot repair an undefined policy or an understaffed follow-up process.
Evidence to require before approval
| Decision area | Evidence to inspect | Acceptance question |
|---|---|---|
| Threat scope | Defined prohibited items, authorized test objects, orientations, concealment conditions, exclusions, and approved procedures | Does the test represent what the facility actually needs to screen? |
| Configuration | Model, software, algorithm version, sensitivity, lane layout, accessories, integrations, and change record | Can the approved result be reproduced after an update or setting change? |
| Alert performance | Expected and observed alerts, misses, nuisance alarms, unresolved events, sample size, and uncertainty | Are limits visible instead of hidden inside a headline percentage? |
| Full-process flow | Arrivals, portal passages, diversions, bag checks, secondary screens, queue length, processing time, and staffing | Can the entire entrance handle peak demand without blocking egress or bypassing policy? |
| Human operation | Role matrix, training results, scenario tests, alert interpretation, supervisor escalation, and relief coverage | Can each operator perform the approved workflow under representative pressure? |
| Accessibility and communication | Accessible route, reasonable-modification process, effective communication, mobility-device and medical-item procedures, and service-animal handling | Can people with disabilities use the process without an improvised or discriminatory barrier? |
| Privacy and data | Notice, images, logs, accounts, permissions, retention, deletion, exports, vendor access, support access, and incident records | Is every collected or displayed data element necessary, protected, and assigned to an owner? |
| Failure and lifecycle | Power, network, display, sensor, integration, and staffing failure tests; inspection; calibration; updates; support; rollback; and replacement | Does the facility have a tested fallback and a named long-term owner? |
Run a representative pilot and acceptance test
- Approve the test plan first. Define safe test procedures, roles, objects, personal items, traffic patterns, environmental conditions, settings, metrics, stop conditions, and decision authority.
- Test the operating range. Include normal and peak flow, representative bags and belongings, operator changes, lane closure, approved sensitivity changes, equipment faults, and the fallback process.
- Measure every disposition. Record what alerted, what did not, what was cleared, what remained unresolved, how long follow-up took, and which staffing or layout condition influenced the result.
- Correct and retest. Track open defects and configuration changes. A successful demonstration is not acceptance if the production workflow, software, environment, or staffing differs.
- Approve the complete package. Require the as-built layout, configuration, account custody, training records, procedures, test results, maintenance plan, support path, privacy controls, and fallback plan.
Accessibility, privacy, and legal review
Screening policy should be reviewed by the organization’s counsel and responsible accessibility, privacy, labor, education, security, and public-safety stakeholders. The Department of Justice Title III regulations require covered public accommodations to address eligibility criteria, reasonable modifications, effective communication, and service-animal access. Coverage and required actions depend on the organization and circumstances, so this guide is not legal advice.
Translate those obligations into an operational plan: accessible queue and passage, clear signage, an equivalent screening path, private follow-up where appropriate, effective communication, trained staff, a documented exception process, and a way to handle mobility devices, medical items, service animals, and other needs without abandoning the security objective.
Connect screening to the rest of the security program
A screening checkpoint may depend on video surveillance, access control, commercial alarm systems, communications, networking, accounts, logging, and response procedures. Define what each integration does, who can access it, which event is recorded, and what happens when it is unavailable. Review network design and cybersecurity ownership for connected systems.
School and municipal teams can pair this technology comparison with Rivell’s school screening planning guide and public-safety technology readiness checklist.
Questions to ask every provider
- Which current model, software version, sensitivity, accessories, and subscription are included?
- Which items and operating conditions were tested, by whom, using what method and sample size?
- Which threats, materials, orientations, concealments, environments, and scenarios are outside the documented scope?
- How do sensitivity changes affect misses, nuisance alarms, secondary screening, queueing, and staffing?
- Which personal items or bags need diversion, divestment, or separate screening?
- What data is collected, displayed, retained, transmitted, exported, or available to the vendor?
- What changes after a software or algorithm update, and how is regression testing performed?
- What are the inspection, calibration, training, support, replacement, exit, and rollback responsibilities?
Rivell can help New Jersey organizations compare the technology with their entry conditions, infrastructure, staffing, secondary-screening process, integrations, testing, and long-term support needs. Request a concealed weapon detection planning review before accepting equipment or performance claims.
