SovGuardAI analyzes legal documents for indicators of sovereign citizen ideology and related pseudolegal frameworks. It identifies tactics, flags phrases, and surfaces applicable court rejection language — giving attorneys, prosecutors, judges, and court staff a structured starting point for document review. It does not render legal conclusions, make determinations about any individual, or substitute for attorney judgment.

No. SovGuardAI is a document analysis tool, not a legal services provider. The scores, tactic identifications, and briefs generated by SovGuardAI are analytical outputs — not legal conclusions, legal opinions, or legal advice. Nothing produced by SovGuardAI constitutes an attorney-client relationship, and no output should be relied upon as a substitute for independent legal judgment. SovGuardAI is designed to assist in identifying ideological patterns in documents. Any attorney of record remains fully responsible for all legal analysis, strategy, and advice provided to their client. The tool flags — you decide. If you have questions about how to respond to a filing flagged by SovGuardAI, consult qualified legal counsel.

SovGuardAI is for anyone who encounters sovereign citizen ideology in documents — legal professionals and the general public alike. Attorneys — criminal defense, civil litigation, and others who need to quickly assess whether a filing reflects sovereign citizen ideology before investing response time. Court clerks and administrators — triaging incoming filings to flag potential sovereign citizen documents for judicial attention. Journalists, researchers, and advocates — analyzing documents for ideological patterns in the course of reporting, research, or monitoring. Individuals and families — anyone who has received a document, notice, lien, or communication and wants to understand whether it reflects sovereign citizen tactics. You do not need to be a legal professional to use SovGuardAI.

The score reflects the density and variety of sovereign citizen pattern matches detected across the document — not a legal finding or prediction of outcome. A higher score reflects greater density and variety of known pattern matches. A lower score reflects their absence — not their impossibility. The score runs from 0 to 100 across four tiers: 0–10 (None): Below detectable limits. No meaningful sovereign citizen pattern detected. 11–40 (Low): Present but not prevalent. Isolated markers present — may reflect copied language or incidental phrasing. Important: a Low score is not a clean bill of health. The taxonomy is actively maintained, but no tool can detect patterns that have not yet been documented. 41–70 (Moderate): Notable ideology present. Multiple tactics detected across the document. Warrants closer review. 71–100 (High): Strong sovereign citizen ideology detected. Consistent with dedicated sovereign citizen filing strategy. For a full explanation of how the score is calculated and what each tier means, see the Methodology page.

SovGuardAI was specifically stress-tested against documents designed to trigger false positives — including academic analysis of sovereign citizen ideology, standard pro se prisoner civil rights complaints with isolated phrasing, and documents that share surface features with sovereign citizen filings but lack ideological intent. Tactic detection operates at the conceptual level, not keyword matching. A document that analyzes sovereign citizen arguments scores near zero. A standard pro se complaint with one or two isolated phrases scores Low. The tool is calibrated to distinguish ideology from incidental similarity. No automated tool eliminates false positives entirely. Review of the specific tactics identified is always recommended before acting on a score.

Court rejection language in every SovGuardAI report is drawn from a fixed reference set — not generated by the model at analysis time. Citations are hardcoded, so the model cannot invent, alter, or hallucinate a case name or reporter cite in your report. This is a deliberate architectural decision: AI-generated legal citations are a known failure point across the industry, and removing citation generation from the model's responsibilities removes that specific failure mode. It does not remove your verification duty, and we do not claim it does. A hardcoded citation can still be inaccurate, superseded, no longer good law, or an imperfect match for the proposition it is cited alongside — and hardcoding cannot detect any of those. Our Terms require you to confirm every citation against Westlaw, Lexis, or CourtListener before relying on it. That requirement is real, not boilerplate.

SovGuardAI is stress-tested against real federal court filings pulled from PACER — not synthetic test documents. The test corpus spans high-density sovereign citizen filings, prisoner civil rights complaints, complex mixed-content scanned documents, and adversarial documents designed to evade detection without using trigger vocabulary. The suite also covers adversarial evasion, false positives, mixed-content documents, and prompt injection. Testing to date indicates that concept-level pattern detection — not keyword matching — is the correct scoring signal, and that tactic density across document sections is more informative than document type or form used. The stress suite is a regression guard: it runs before any prompt change is deployed and flags when behavior shifts against documents whose expected result has already been set. It is not a measurement of accuracy. Precision and recall figures require human-adjudicated ground truth — a labeled corpus in which qualified reviewers, rather than the model, establish the correct answer for each document — which the suite does not provide. All testing described here is internal. The same document may score within a range across separate analyses. The score is intended as triage — a prompt to look closer — rather than a measurement.

Your results are available for 7 days via a secure link. Your original file is never stored — the text extracted from it and your analysis report are retained for 7 days to deliver your results, then permanently deleted. Nothing you submit is indexed, used for training, or shared with third parties. Beyond that window we keep anonymous statistics about which tactics appear in scanned documents — never the documents themselves, and never anything that identifies a person or case. If you have specific confidentiality requirements or are subject to professional rules governing client document handling, review our Privacy Policy before submitting sensitive materials.

Whether submitting client documents to a third-party AI tool is permissible depends on your jurisdiction's professional responsibility rules, your engagement agreement, and the nature of the material. Results are available for 7 days via a secure link. Your original file is never stored; the extracted text and your report are retained for 7 days, then permanently deleted. Beyond that window we retain only de-identified statistics about completed analyses — never document contents, names, or case numbers. Attorneys should independently assess their obligations before uploading client materials. When in doubt, consider redacting identifying information before submission, or consult your jurisdiction's ethics guidance on use of AI tools with client data.

Yes. SovGuardAI processes scanned PDFs and handwritten filings, including documents with annotations, stamps, and mixed content. Detection quality is affected by document quality: native text PDFs produce the most reliable results, and heavily scanned or handwritten documents may limit phrase extraction, particularly where key language appears in degraded form. Scoring and tactic detection draw on pattern recognition across the document rather than exact phrase matching, so they tolerate a degree of OCR noise — but a badly degraded scan can still cause missed detections. Where OCR degradation is detected, SovGuardAI notes it in the analysis and recommends manual review of the original filing. Documents identified as scanned will show a banner noting this.

SovGuardAI is priced at $59 per scan. Each scan includes the full risk score, tactic breakdown, and 60-second brief for one document. Have questions about volume pricing or organizational access? Contact us at support@sovguardai.ai.