Built for the Courtroom: The StreetGuard Evidence Mechanism
How tamper-evident hashing, trusted timestamps, and AI-validated geofencing turn a 30-second citizen video into a citation that holds up in court.
6 min read
The evidence problem in parking enforcement
A parking citation is only as strong as the evidence behind it. When a driver contests a fine, the city must prove — to a hearing officer or a traffic judge — exactly what happened, exactly when it happened, and that the record has not been altered since it was created. In the era of traditional officer-issued citations, that proof rested on the officer's sworn statement and a handwritten ticket. In an AI-assisted, citizen-reported enforcement system, the standard is higher: digital evidence is inherently suspect unless it carries a verifiable, unbroken chain of custody from the moment of capture to the moment of adjudication.
StreetGuard was designed from the ground up to meet — and exceed — that standard. The platform's evidence mechanism is not a compliance checkbox. It is the architectural foundation on which every other feature rests. This paper describes how it works, why each component matters, and what it means for a city's ability to issue citations that hold up in court.
StreetGuard turns a 30-second video from any phone into a court-defensible parking citation — with tamper-evident evidence, AI license plate reading, and one-click officer review.
The four-stage evidence pipeline
Every citation generated by StreetGuard passes through four mandatory stages. No stage can be skipped. No stage can be reversed. Each stage adds a layer of verifiability that compounds the integrity of the final evidence packet.
- 1
Citizen records
A resident opens StreetGuard in any phone browser — no app download required — and records a 30-second clip of the violation. GPS coordinates are captured automatically at the moment of recording, and the platform detects the city jurisdiction from the device's location. The simplicity is intentional: the lower the friction for the reporter, the higher the volume of actionable evidence the city receives.
- 2
Evidence sealed — before anyone reviews it
This is the most critical stage, and it happens automatically before any human sees the footage. The video is cryptographically hashed (SHA-256), timestamped via an external RFC 3161 Trusted Stamp Authority, and signed into a tamper-evident manifest. The RFC 3161 timestamp is issued by a third-party authority the platform does not control — meaning the capture time is independently verifiable by any party, including a defence attorney or a court. The original media is then written to WORM (Write Once, Read Many) storage. Once sealed, the evidence cannot be modified by anyone inside or outside the platform.
- 3
AI detection and geofence validation
With the evidence sealed, the AI pipeline runs. A multi-modal model reads the license plate with character-level confidence scoring, classifies the violation type, and cross-references the location against the city's configured geofence zones — fire-hydrant boundaries, no-stopping windows, permit exceptions, and time-of-day rules. Critically, the AI operates on the already-sealed evidence: it cannot alter what was recorded, only interpret it. Every inference output is logged with the model version that produced it, making every AI decision reproducible and auditable.
- 4
Officer approval and citation generation
A deputised city officer reviews the AI's assessment in the review queue. They see the original video, the AI detection trace, the applicable rule, and the plate read. One click approves the citation; one click rejects it. Every officer action is logged with the officer's identity, session attestation token, IP address, and timestamp. On approval, StreetGuard generates a serial-numbered citation PDF that embeds the evidence ID, the SHA-256 hash, the capture coordinates, and the officer's name — a document that can be printed, mailed, and defended in any jurisdiction.
Six features that make evidence court-ready
Tamper-evident hash chain
Each city's audit log is a cryptographic chain: every event row hashes the previous row's hash together with its own payload. Removing or modifying any audit event breaks the chain — and the break is provable. A defence attorney who claims the evidence was altered must contend with a mathematical proof that it was not. The chain is per-tenant: City A's evidence chain has no connection to City B's, eliminating any theoretical cross-contamination argument.
RFC 3161 trusted timestamps
The RFC 3161 standard is the international benchmark for trusted digital timestamps. When StreetGuard seals an evidence item, it requests a timestamp token from an external Timestamp Authority — a service that is cryptographically independent of StreetGuard. The token binds the evidence hash to a wall-clock time that StreetGuard did not control and cannot retroactively change. This directly defeats the most common digital evidence defence: 'the timestamp was set by the same party that produced the evidence.'
Per-tenant isolation
Each city's evidence lives in its own isolated namespace — its own MinIO storage bucket, its own audit chain, its own data retention policy. There is no shared data store that could create arguments about commingling of evidence from different jurisdictions. Cities can independently configure retention windows: cited evidence kept for years for litigation readiness; no-violation clips purged on a 14-day cycle to minimise storage cost and privacy surface area.
Court-ready citation PDFs
The citation document generated at the end of the pipeline is not a generic printout. It embeds the evidence ID (a globally unique identifier that locates the original media), the SHA-256 hash of that media (so the court can verify the exhibit matches the original), the GPS coordinates of the capture, the applicable regulation, the AI confidence score, and the reviewing officer's name and badge. Everything a hearing officer or magistrate needs to evaluate the citation is present on the face of the document.
Per-city configurable rule zones
Evidence reliability depends not just on what was captured but on whether the applicable rule was correctly identified. StreetGuard's interactive geofence map allows each city to define fire-hydrant clearance zones, no-stopping windows, permit-holder exceptions, and time-of-day schedules with point-and-click precision. Rule sets are versioned: when the AI resolves whether a violation occurred, it uses the rule version that was active at the evidence timestamp — not the current version. This ensures that a rule change after the fact cannot retroactively invalidate or validate past citations.
Encrypted transit and storage
Every video submitted through the citizen reporter is encrypted in transit (TLS 1.3) from the device to StreetGuard's ingestion gateway. Original media is stored encrypted at rest under per-tenant keys. Access to evidence objects is mediated through short-lived signed URLs — no long-lived credential holds direct object access. The encryption architecture means that even in the event of a storage-layer breach, raw evidence files are unreadable without the per-tenant key material.
What this means for cities in practice
The practical implication of StreetGuard's evidence architecture is a significant reduction in the cost and risk of contested citations. In a traditional enforcement system, a contested citation typically requires an officer to appear at a hearing and testify. With StreetGuard, the evidence packet is self-documenting: the SHA-256 hash, the RFC 3161 timestamp, the officer's logged approval action, and the court-ready PDF collectively constitute a record that can be submitted to an adjudication system or presented to a hearing officer without requiring the reviewing officer to appear in person.
Cities that have deployed video-based enforcement consistently report that dispute overturn rates fall sharply once defendants — and their attorneys — understand the evidentiary standard they are contesting. StreetGuard targets a dispute overturn rate below 5 percent, a figure that reflects not just AI detection accuracy but the depth of the evidence record supporting each citation. A driver who argues 'that wasn't my car' faces a SHA-256-verified, RFC 3161-timestamped video of their plate. A driver who argues 'the rule didn't apply there at that time' faces a versioned geofence record showing exactly what the rule was at that location and timestamp.
If a driver contests, you can show exactly what was captured, when, and that nothing was modified.
Beyond individual citations, the tamper-evident audit chain creates an institutional record that supports broader accountability. City managers, auditors, and oversight bodies can verify — independently, without relying on StreetGuard's representations — that the enforcement system operated as configured, that no evidence was suppressed or altered, and that the officer review process was followed for every citation issued. This level of transparency is increasingly required by municipal procurement frameworks for AI-assisted enforcement systems, and StreetGuard is built to meet it from day one.
Conclusion: evidence as infrastructure
StreetGuard's evidence mechanism reflects a design philosophy that most enforcement technology platforms get backwards: evidence integrity is not a feature to be added after the core product is built. It is the core product. Every architectural decision — the RFC 3161 third-party timestamp, the WORM storage, the per-tenant hash chain, the versioned rule sets, the officer action logging — exists to answer a single question that every contested citation will eventually ask: how do we know this evidence is real, unaltered, and correctly interpreted?
Cities that deploy StreetGuard are not just acquiring an enforcement tool. They are building a public record of their streets — one that is cryptographically verifiable, legally defensible, and independently auditable. That record is the foundation on which safe streets, reliable revenue, and trustworthy AI governance are built.
See a violation? Report it now.
Takes thirty seconds. Your camera, our pipeline, your city's enforcement.
Report a parking issue