Fargo, West Fargo & Cass CountyCriminal defense informationCall 701-929-4886
Fargo and North Dakota criminal defense guide

Automated License-Plate Reader Evidence in North Dakota

An automated license-plate reader photographs a vehicle, uses software to propose a plate number, and stores the event with time and location information.…

ElementsPenaltiesEvidenceNext steps

Fixed and mobile ALPR cameras, optical character recognition, plate images, context photographs, GPS, timestamps, hot lists, alerts, historical searches, database retention, vehicle and driver attribution, traffic stops, discovery, experts, and suppression

Automated License-Plate Reader Evidence in North Dakota

An automated license-plate reader photographs a vehicle, uses software to propose a plate number, and stores the event with time and location information. It may compare the proposed number against a hot list or preserve the scan for later searching. An alert is an investigative lead, not proof that the software read the plate correctly, that the hot-list record was current, that the observed vehicle was the wanted vehicle, or that its registered owner was driving. Defense review should begin with the original images, OCR alternatives, camera and clock data, list source, alert history, officer confirmation, stop chronology, database query, and retention record.

An ALPR hit is a computer-generated hypothesis. Verify the plate image, state and plate type, confidence and alternate reads, context image, time zone, camera location, hot-list entry and update time, officer’s visual confirmation, vehicle description, plate status, and every later database query.

An ALPR record combines an image, software output, and metadata

The camera may be fixed, mobile, or privately operated

Readers can be mounted on poles, roads, parking facilities, trailers, patrol vehicles, tow trucks, or private security systems. A fixed camera records vehicles passing one point; a mobile camera records wherever the host vehicle travels. Law-enforcement databases may combine agency-owned scans with data shared by other agencies or vendors. Source ownership and access history matter.

Optical character recognition proposes the plate text

The software detects a plate-shaped region, separates characters, and calculates a likely alphanumeric string. Dirt, snow, glare, angle, speed, damage, frames, temporary tags, specialty plates, stacked characters, and similar symbols can cause errors. The proposed text should be compared with the plate image, not accepted as though the camera directly received registration data.

A plate image and context image serve different purposes

The cropped plate image supports the character read. A wider context image may show vehicle type, color, lane, direction, occupants, surroundings, or another nearby plate. Compression and exposure may affect each differently. A correct text read attached to the wrong context image, or a context image containing two vehicles, can create attribution problems.

The event record usually includes time and camera location

Metadata may include UTC or local time, GPS coordinates, camera ID, direction, lane, confidence, alternative reads, operator, source agency, and software version. A visible overlay can be added later. The raw stored value, time zone, device or server clock, synchronization, and camera coordinates should be preserved.

Hot lists come from separate source systems

An ALPR system may compare proposed plate text against lists for stolen vehicles, wanted persons, missing persons, revoked registrations, investigative interests, or other authorized purposes. The list source controls what the alert means. Update frequency, entry criteria, expiration, cancellation, and matching rules affect reliability. The camera does not independently know a vehicle is stolen.

Fuzzy matching can alert on nonidentical characters

Systems may generate alerts when most characters match or when common OCR confusions occur. This can catch imperfect images but also produce false positives. Discovery should reveal match rules, wildcard use, confidence, alternative reads, and whether the alert was exact. Officers should view the image and plate before acting when circumstances permit.

A scan can be stored even when no alert occurs

Historical databases may retain plate, time, camera, location, image, and query history for ordinary vehicles that generated no investigative interest at collection. Retention periods, sharing, audit logs, access permissions, and vendor agreements determine how that information can later be used. A real-time hot-list system and a searchable movement database raise different issues.

A real-time alert must be reliable when the vehicle is stopped

The officer should confirm plate, state, and vehicle

An alert may display proposed text and an image. The officer can compare the observed plate character by character, identify issuing state and plate type, and compare make, model, color, year, and other registration information. Failure to notice an obvious mismatch can make reliance unreasonable. A dangerous situation may affect how much confirmation is feasible, but it does not correct the underlying record.

The alert’s source determines the suspected offense

“Hit” is not a legal category. A stolen-vehicle entry, wanted-person association, expired registration, investigative bulletin, or private repossession interest carries different authority. The State should identify the exact source record, entry language, entering agency, date, validation, and status at the stop.

Hot-list latency can make a once-valid alert stale

A recovered vehicle, cleared warrant, renewed registration, or corrected plate may remain in a local cache until the next update. Determine when the source system changed, when the ALPR system last synchronized, when the patrol device received the update, and whether officers had contrary information. Good-faith reliance and the existence of reasonable suspicion are fact specific.

A plate is associated with a vehicle before it is associated with a driver

Registration can identify the owner, not necessarily the current driver. Vehicles are loaned, shared, rented, sold, and stolen. Plates can be transferred, cloned, altered, or mounted on the wrong vehicle. A driver-identification inference should consider vehicle description, officer observation, license photo, occupants, possession, and information learned before the stop.

Glover permits an inference but preserves contrary facts

In Glover, the Supreme Court held that the deputy’s inference that the registered owner was driving was reasonable under the totality of circumstances when the owner’s license was revoked and the deputy lacked information negating it. The Court emphasized the modest reasonable-suspicion standard and case-specific commonsense reasoning. A visible mismatch in age, sex, vehicle, or driver description can change the analysis.

The stop cannot be prolonged beyond its lawful mission without additional cause

Rodriguez v. United States, 575 U.S. 348, 354–57 (2015), holds that police may not prolong a traffic stop beyond the time reasonably required for its traffic mission to conduct unrelated investigation without independent reasonable suspicion. An ALPR alert can justify the initial seizure without authorizing unlimited questioning, waiting, dog deployment, or device searches.

A mistaken alert and a mistaken officer inference are separate issues

The OCR may be wrong, the hot list may be wrong, or the officer may misread a correct alert. Each error has a different source, custodian, audit record, and legal analysis. The defense should identify exactly which step failed and what the officer knew when acting.

Collective knowledge requires a reliable originating record

An officer may act on an alert or bulletin entered by another officer or agency without personally knowing every underlying fact. The State should still identify the source agency, person or system responsible for entry, factual basis, permitted purpose, validation, and communication chain. Passing a conclusory or unsupported entry through an automated database does not improve the information’s reliability merely because the stopping officer received it electronically.

Cancellation and dissipation can end the basis for detention

If dispatch reports that a vehicle was recovered, a warrant was cleared, the plate was entered in error, or the observed vehicle does not match, reasonable suspicion may dissipate. The chronology should include every radio transmission, terminal update, officer acknowledgment, and observation. Officers may address safety and confirm ambiguity, but they cannot ignore definitive contrary information and continue an investigative detention on the original premise.

A vehicle-description conflict may be more significant than an exact plate read

An exact OCR result associated with a different make, model, color, or body style can indicate a stolen, cloned, swapped, or incorrectly assigned plate. The conflict may justify investigation of plate misuse, but it weakens an inference that the observed vehicle is the wanted vehicle described in the hot list. The State should disclose historical registration, plate status, source description, and what the officer saw before the seizure.

Historical ALPR queries can reconstruct vehicle movement

A single scan proves only a camera observation within its limits

A reliable record can show that a plate image was captured near a camera at a time. It does not prove the registered owner occupied the vehicle, that the vehicle stopped, or where it traveled next. Camera coordinates, lane, direction, clock, image, and plate match should be verified.

Repeated scans can create a route and behavioral pattern

Multiple records may suggest travel between points, presence near a residence, work, clinic, worship site, political meeting, or association. Query range, camera coverage, missing areas, plate changes, and false matches affect the pattern. A database is not a continuous GPS track; it samples only where participating cameras captured a readable plate.

Search criteria can be broader than an exact known plate

Systems may support partial plates, wildcard characters, vehicle descriptions, geographic boxes, time windows, common-location reports, convoy analysis, or lists of every vehicle at a place. Each technique has a different false-positive population and privacy effect. Request the actual query, filters, results, excluded records, and analyst steps.

Common-plate and common-location searches require careful validation

Comparing vehicles seen at several scenes can produce investigative leads. Popular routes, dense parking, delivery vehicles, commuters, fleet plates, OCR errors, and long time windows generate innocent matches. The method should disclose the complete population, number of scenes, time tolerances, plate normalization, and ranking criteria.

Private and interagency data can expand coverage beyond the local system

A local investigator may query scans collected by another agency, a national vendor, parking operator, repossession company, neighborhood network, or border system. Determine who collected the record, governing agreement, permitted purpose, retention, quality controls, access, and whether local audit logs capture the external query.

Retention and deletion affect both privacy and defense access

Short retention limits historical reconstruction; long retention creates a deeper movement archive. Relevant records may expire after the prosecution exports selected hits. Defense counsel should promptly seek preservation of source scans, nonmatching nearby records, query logs, configuration, and audit data. A printed map is not a substitute for the underlying database return.

Query history can reveal investigative evolution and confirmation bias

Audit logs may show when a plate was first searched, what ranges were used, who viewed results, how criteria changed, and which plates were excluded. Later knowledge can influence interpretation of earlier ambiguous images. The defense should distinguish the original automated result from analyst refinement after a suspect emerged.

ALPR evidence has technical, geographic, and attribution limits

OCR confidence is not a universal probability of correctness

A confidence score depends on vendor model, training, plate type, image quality, and calibration. It may rank alternatives without representing a validated error probability for North Dakota plates. The witness should explain what the score means and how performance was tested on comparable images.

Similar characters and plate formats create predictable errors

Zero and O, one and I, five and S, eight and B, temporary plates, stacked characters, specialty designs, frames, dirt, and snow can confuse recognition. State and plate-type classification can also be wrong. Review all captured frames and alternative reads rather than only the database’s final text.

Camera location can be misstated

A fixed camera may be mapped to a pole, intersection centroid, or approximate coordinate. A mobile reader’s GPS can lag or be unavailable. Trailer cameras can move. Calibration, installation records, GPS logs, and maintenance history establish location. A map should not show greater precision than the source.

Timestamp errors can reverse a travel theory

Device clocks, servers, GPS time, UTC conversion, daylight saving, delayed uploads, and mobile-system settings can differ. Preserve raw times and offsets. Compare known events and neighboring scans. A seconds-level travel calculation is unreliable if cameras differ by minutes.

The plate may not belong on the observed vehicle

Cloned, stolen, swapped, expired, obscured, or counterfeit plates can produce a technically correct read linked to the wrong registration. Compare vehicle make, model, color, body style, damage, decals, and registration history. A plate-only match should not erase a clear vehicle mismatch.

One vehicle can have multiple plates or temporary identifiers over time

Plate replacement, transfer, temporary permits, dealer plates, fleet registration, and jurisdiction changes complicate historical searches. A current plate-to-vehicle relationship may not apply on the scan date. Registration records should be historical, not merely current.

The database can contain duplicates and derivative records

Several frames from one pass, shared copies across systems, repeated imports, and reprocessed OCR can appear as multiple events. Deduplication rules should be disclosed. Apparent corroboration is weak if several entries derive from the same camera capture.

North Dakota and federal law distinguish a visible plate from prolonged tracking

A plate displayed on a public road is ordinarily exposed to observation

Officers can generally observe and check a visible license plate while lawfully present. Delaware v. Prouse, 440 U.S. 648, 653–63 (1979), nevertheless rejects random discretionary vehicle stops merely to check license and registration. Automated observation does not eliminate the requirement for an objective basis to seize the vehicle.

North Dakota’s unmanned-vehicle statute excludes license-plate readers from “robot”

N.D.C.C. § 29-29.4-01(8) excludes a law-enforcement body camera, in-car camera, and license-plate reader from that chapter’s definition of “robot.” That definitional exclusion is not a comprehensive ALPR evidence code and does not resolve every constitutional question about a particular stop, database, query, or prolonged tracking practice.

Jones addresses physical GPS installation, not ordinary camera capture

United States v. Jones, 565 U.S. 400, 404–13 (2012), held that attaching a GPS device to a vehicle and using it to monitor movement was a search because of physical trespass. ALPR cameras do not ordinarily attach a device to the car. The concurring opinions nevertheless discuss privacy concerns from long-term comprehensive movement monitoring.

Carpenter protects comprehensive location history but is narrow

Carpenter v. United States, 585 U.S. 296, 309–22 (2018), held that acquiring at least seven days of historical CSLI was a search. The Court emphasized detailed, encyclopedic, easily compiled movement history while stating that its decision was narrow and did not call conventional security cameras into question. It did not decide ALPR databases.

ALPR constitutional analysis should match the actual use

A single real-time plate check, a short query around a crime scene, a months-long search of one vehicle, and a reverse search of every vehicle near sensitive places are materially different. Courts should examine duration, density, database breadth, public and private sources, retrospective reach, automation, purpose, and whether the query reveals a comprehensive movement pattern.

A database hit does not itself authorize a search of occupants or devices

The alert may justify stopping the vehicle or investigating its status. A search of the passenger compartment, trunk, occupants, or phones requires consent, warrant, probable cause, search-incident limits, inventory authority, or another recognized doctrine. The scope of the stop should not expand merely because the lead was technologically generated.

Affidavits must disclose material ALPR limitations

A warrant application relying on scans should identify original images, exact and alternative reads, time conversion, camera, database source, vehicle consistency, retention, and query method. Material falsehoods or omissions concerning misreads, mismatched vehicles, stale lists, or location can support a request under Franks v. Delaware, 438 U.S. 154, 155–56, 171–72 (1978).

ALPR records require authentication and reliable explanation

The proponent must identify what the record claims to be

Under N.D.R.Ev. 901, the State may authenticate images, metadata, hot-list records, database returns, and maps through witnesses with knowledge, distinctive characteristics, and evidence about accurate systems. A patrol officer may explain receiving an alert without knowing the vendor’s OCR, retention, or historical-query process.

Electronic certifications can establish copy or system integrity

N.D.R.Ev. 902(13) and (14) allow qualifying certified electronic-process records and copied electronic data, with notice and inspection. Certification does not establish that OCR was correct, the driver was the owner, or an analyst’s route inference is valid.

Machine output and human-entered hot-list data are different

The camera and OCR can generate an image, plate proposal, and metadata. An officer or agency may enter the hot-list record and investigative note. Each human statement may require its own hearsay basis. A business-record foundation for the ALPR database does not automatically prove the truth of an embedded accusation.

Expert testimony may be needed for disputed methodology

N.D.R.Ev. 702 applies to specialized opinions about OCR performance, camera location, timing, route reconstruction, common-location analysis, photogrammetry, or database completeness. Disclose data, model or version where available, match rules, queries, validation, error limits, exclusions, and reproducible calculations.

Maps and timelines can overstate continuity

A line drawn between sparse scans implies a route the database did not observe. Equal dots may represent different confidence or source systems. A fair exhibit identifies every actual scan, gaps, camera coverage, time uncertainty, and alternative routes. N.D.R.Ev. 107 governs illustrative aids, while Rule 1006 governs evidentiary summaries of voluminous admissible material.

Rule 403 guards against algorithmic overstatement

A jury may give a computer label undue weight. N.D.R.Ev. 403 permits exclusion when misleading effect or unfair prejudice substantially outweighs probative value. Accurate terminology—“proposed read,” “alert,” “registered owner,” and “scan”—helps preserve distinctions the evidence actually supports.

Defense counsel should reconstruct the alert and every historical query

Obtain native records from each system

Request original plate and context images, all frames, raw OCR values, alternatives, confidence, timestamps, time zones, camera ID and coordinates, GPS logs, calibration and maintenance, software version, hot-list entry, updates and validation, alerts, officer display, query records, audit logs, retention, sharing agreements, vendor exports, maps, and derivative warrants.

Build a scan-to-stop chronology

Record image capture, OCR, list comparison, alert creation, transmission, officer receipt, visual confirmation, database checks, observations, seizure, questioning, search, arrest, and later correction. The chronology identifies whether reasonable suspicion existed at the moment of the stop and whether it dissipated.

Compare plate and vehicle independently

Inspect state, plate type, every character, temporary or specialty formatting, vehicle make, model, color, body style, damage, decals, occupants, and historical registration. Obtain alternate frames and enhancement history. A technically readable plate can be stolen or cloned; a correct vehicle description cannot cure a different plate.

Audit historical-query methodology

Demand exact query syntax, date and geography, wildcards, normalization, vehicle filters, common-location thresholds, excluded results, deduplication, ranking, analyst notes, and the complete result population. Reproduce the search. Determine whether investigators changed criteria after learning the suspect’s plate.

Preserve exculpatory nonmatches and surrounding scans

Records from other cameras may show a different route, impossible travel, another vehicle with a similar plate, or absence where coverage was reliable. Absence must be interpreted carefully because no scan can result from traffic lane, occlusion, image quality, or camera outage. Source health records and coverage are essential.

Use experts for the actual disputed layer

An image examiner can analyze plate characters; a network or system expert can review clocks and databases; a statistician may evaluate common-location matching; a photogrammetry expert may address vehicle dimensions or speed. The assignment should be narrow, reproducible, and based on native evidence.

File issue-specific motions

Potential issues include lack of reasonable suspicion, stale or mismatched alert, prolonged detention, unlawful search scope, historical-query privacy, warrant nexus, material affidavit error, missing discovery, unauthenticated records, hearsay hot-list entries, unreliable expert method, and misleading demonstratives. Request suppression, a Franks hearing, native production, continuance, Rule 702 hearing, correction, limiting instruction, or exclusion tailored to the record.

Separate vehicle presence from criminal participation

Even a reliable scan proves only that the observed plate and vehicle passed a camera at a time. It does not identify the driver, occupants, purpose, destination, knowledge, or conduct outside the frame. Registration, possession, surveillance, device data, witnesses, and other evidence must establish the human link and each charged element.

Frequently asked questions about license-plate-reader evidence

What is an automated license-plate reader?

It is a camera system that photographs vehicles, uses optical character recognition to propose plate text, records metadata, and may compare the result with a hot list.

Can ALPR software misread a plate?

Yes. Angle, glare, speed, snow, dirt, plate design, similar characters, frames, and temporary tags can cause errors. The original image and alternative reads should be reviewed.

Does an ALPR hit justify a traffic stop?

A reliable current hit can support reasonable suspicion or probable cause depending on its meaning and circumstances. Officers should confirm plate and vehicle when feasible.

Does a plate identify the driver?

No. It identifies a registration relationship to a vehicle. The owner may not be driving, and the plate may be transferred, stolen, cloned, or mounted on another vehicle.

What did Kansas v. Glover decide?

It allowed a commonsense inference that a revoked registered owner was driving when the officer lacked facts negating the inference. It did not make owner identity conclusive.

What is a hot list?

It is a source list against which proposed plate reads are compared, such as stolen vehicles or wanted-person associations. Entry, update, and removal reliability matter.

Can a stale alert invalidate a stop?

It can undermine reasonable suspicion depending on what officers knew and whether reliance was objectively reasonable. Source updates and synchronization should be examined.

Do ALPR systems store non-hit plates?

Many systems can store ordinary scans for later searching, subject to agency, vendor, contract, and retention rules. The actual system and policy must be verified.

Is one ALPR scan the same as GPS tracking?

No. One scan is a camera observation at a point. Repeated scans can create movement history, but coverage remains intermittent and legal analysis depends on scope and duration.

Has the Supreme Court decided historical ALPR database searches?

No. Jones and Carpenter address related tracking and aggregation concerns but did not decide ALPR databases.

Does North Dakota have a comprehensive ALPR evidence statute?

N.D.C.C. ch. 29-29.4 excludes license-plate readers from its definition of robot, but that provision does not resolve every stop, database, retention, privacy, or evidentiary issue.

Can ALPR locate a person?

It can record a vehicle plate at a camera. It does not establish who occupied the vehicle or where any person went after the scan.

Why are raw timestamps important?

Clock, time-zone, GPS, server, upload, and daylight-saving differences can alter a chronology. Raw values permit independent conversion and comparison.

What is a common-location search?

It compares vehicles observed at multiple places or events. Complete populations, time windows, false positives, normalization, and ranking criteria require review.

How are ALPR records authenticated?

Through witnesses, certifications, source images, metadata, system evidence, audit logs, and preserved database records showing what the exhibit claims to be.

When is an expert needed?

An expert may be useful for OCR, image analysis, timing, camera location, database completeness, route reconstruction, common-location analysis, or map accuracy.

What should defense counsel request?

All source images and frames, raw OCR and alternatives, timestamps, camera records, hot-list history, alerts, officer displays, queries, audit logs, retention, software versions, and maps.

Can an ALPR alert authorize a vehicle or phone search?

Not by itself. A further search requires consent, warrant, probable cause, search-incident authority, inventory rules, or another recognized basis and must remain within its scope.

Official authorities cited

Legal information, not legal advice. ALPR evidence depends on source images, OCR and alternatives, plate type, timestamps, camera location, hot-list status, officer confirmation, vehicle consistency, database scope, retention, query method, audit history, driver attribution, expert foundation, stop chronology, search scope, and current law.


Call nowChoose a time