Guard the boundary of 3D printing.

PrintGuard identifies firearm-related risk before a file is published, queued, or fabricated—locally, quickly, and with an auditable record.

Local screening Auditable decisions Practical compliance Scroll to explore
Local screening, not central surveillance Geometry over guesswork

A software safety layer between every file and the printer.

PrintGuard can run as a cloud service called through an API or directly on printer hardware. Either way, the same classifier screens each file before fabrication and only releases jobs that clear the configured firearm-part policy.

PG / DEPLOYMENT PATH Screening gate · before motion
01 / Intercept

File queued to print

The safety check starts before the job reaches the fabrication controller.

print_job.3mf

02 / Choose deployment

Cloud service

Call through the API

Slicers, marketplaces, and print services send the file to a managed screening endpoint.

POST /v1/classify
or
Local runtime

Run on printer hardware

The classifier executes on the printer or local controller without sending the file to the cloud.

LOCAL / NO UPLOAD
03 / Classify geometry

Check for firearm-part risk

The classifier reads the shape itself—not the filename—and scores the file before any print motion is authorized.

MeshPointsVoxelsViews

04 / Enforce before print

Cleared

Release to printer

The job continues only after it clears the active classification and policy line.

Firearm-part signal

Apply configured action

Hold, block, or route according to jurisdiction-specific rules and workflow.

INPUT / 01

Hardened file intake

Bounded parsers validate STL, OBJ, AMF, 3MF, and OFF files before analysis.

SIGNAL / 02

Geometry, not filenames

Mesh statistics, point clouds, voxel occupancy, and multiview projections capture the durable shape signal.

RECORD / 03

Auditable policy action

Scores, evidence, thresholds, and the resulting pre-print action are retained in a decision record.

A geometry problem cannot be solved by a list of names.

A

Filenames are fragile

Names can be changed in seconds. Keyword filters confuse intent with a label.

B

Hashes only match exact files

A small file edit produces a new hash, even when the underlying geometry is effectively unchanged.

C

Geometry is the durable signal

PrintGuard extracts mesh statistics, sampled points, voxel occupancy, and multiview projections from the file itself.

Safety that works at the speed of a workflow.

A source-separated test set of firearm designs, firearm-specific parts, dual-use parts, near-firearm duplicates, and benign 3D files.

Firearm detection 96.75%

Percentage of firearm files successfully flagged.

False-positive rate 0.06%

Benign files incorrectly flagged. Lower means less friction for lawful use.

Per-file scoring <1 sec

Per file, for average file sizes.

Held-out evaluation 10,000 files

Source-separated test set.

V1 benchmark · July 2026

Evidence before enforcement.

PrintGuard is built around documented datasets, source-separated testing, clear audit records, and human review.

01 / Training foundation113,368

mesh rows in the binary training manifest

02 / Lawful lookalikes99,848

benign mechanical examples selected to challenge the classifier

03 / Restricted evidence13,520

positive mesh examples from authorized channels

Local-first

Screening can happen where files already live, without centralizing every design.

Auditable by design

Hashes, model scores, evidence, and decisions create a reviewable record.

Human where it matters

Ambiguous and high-impact decisions can move into a review workflow.

Built to evolve

Models, benchmarks, and policy rules can be updated as evidence improves.

A national question, moving state by state.

PrintGuard tracks 2025–26 measures that expressly address printer blocking, 3D firearm manufacture, digital instructions, or serialization, together with selected related laws already in force.

Enacted Active Concluded

Scope: 2025–26 state measures expressly addressing printer blocking, 3D firearm manufacture, digital instructions, or serialization, together with selected related laws already in force. Map color reflects the most advanced listed measure. This focused tracker does not attempt to catalog every general firearm or ghost-gun bill. Each linked measure has substantive language within this scope; bills that merely repeat an existing definition are excluded. It is informational, may not be exhaustive, and is not legal advice. Statuses last reviewed August 12, 2026.

A practical middle path

Make 3D printing safer.
Keep innovation open.

We are building a practical screening layer for additive-manufacturing workflows—designed to reduce risk without adding unnecessary friction for lawful use.

PrintGuard v1 mini · live browser demo

Private demo

PrintGuard v1 mini

This page is not linked from the site. Enter the access phrase to run the classifier on your own files.

The phrase gates this page in the browser only. It is not a security boundary — treat the link as confidential.