HomeCase studies › Case 05
Case 05 · Court admissibility

Every commercial AI detector fails Daubert. Ours doesn’t.

Federal Rule of Evidence 702 requires known error rate + reproducibility. Only signed, deterministic detectors clear that bar.

The failure · on the record

Federal Rule of Evidence 702 sets the Daubert reliability standard: testable methodology, peer-reviewed, known error rate, general acceptance in the field. Proposed FRE 707 (2026) applies these criteria specifically to machine-generated evidence. LLM detectors that update silently do not have a known error rate at any single point in time — their score is a moving target. We publish signed benchmarks that pin our error rate to a specific detector version, verifiable years later.

The numbers
Daubert factorGPTZeroTurnitinBinocularsTruth-in-AI
Testable methodologyPartialNoYesYes (signed artifact)
Peer-reviewedNoNoYesYes (RAID benchmark)
Known error rateNo (drifts)No (drifts)Yes (pinned weights)Yes (signed, per-config)
ReproducibilityNoNoOnly if weights pinnedYes (Ed25519 replay)
General acceptanceCommercialCommercialAcademicNew — growing
What we do differently

Every signed result includes an admissibility packet: the detector’s signed benchmark AUC, its version SHA, the input text’s SHA-256, an Ed25519 signature over all of the above, and a Bitcoin OpenTimestamps anchor.

Enough for a paralegal to file. Enough for a Daubert hearing.

Proof
5 / 5
Daubert factors met · competitors: GPTZero: 1/5 · Turnitin: 0/5 · Binoculars: 3/5
Try it yourself
Request an admissibility packet.
curl -X POST https://api.trigeochiral.com/detect \
  -H 'Authorization: Bearer $KEY' \
  -d '{"text":"...","packet":"admissibility"}'
Bounty · put money where the claim is
Court witness fee
For qualifying cases where a signed Truth-in-AI result was excluded, TriGeoChiral covers reasonable expert-witness fees to defend the result under Daubert.
Sources