Family01
Family Court
Photographs, screenshots and videos tendered in custody and parenting matters where authenticity is contested.
- Custody
- Screenshots
When the photo lies,
the analysis won't.
Griffin is the AI authentication layer we run on contested images and videos. Deepfake detection, GAN-image identification and provenance tracking — court-grade reporting on whether a piece of media is what it claims to be.
- Class
- AI authentication
- Targets
- Deepfake · GAN · edits
- Output
- Authentication report
What We Examine
What Griffin
What Griffin
actually tests.
Synthetic media has moved from novelty to evidentiary problem. Griffin gives us a structured way to interrogate it — without resting a case on a single signal.
Layer01
AI detection engine
Models trained on millions of authentic and synthetic images, calibrated for false-positive control suitable for court use.
Layer02
Deepfake identification
Face-swap, lip-sync and full-body manipulation detection across video frames with confidence reporting.
Layer03
GAN-image identification
Detection of imagery from Stable Diffusion, Midjourney, DALL-E and other generative tools.
Layer04
Tool fingerprinting
Analysis of generation artefacts and tool signatures left in synthetic content.
Layer05
Provenance & edits
C2PA where present, EXIF analysis, recompression history and edit-trail reconstruction.
Layer06
Audio authentication
Voice-cloning detection on contested call recordings and audio attachments.
Coverage
Where Griffin
Where Griffin
enters a brief.
Most calls for AI authentication arrive from one of four directions — and each requires a different threshold of confidence.
Criminal02
Criminal defence
Defendant-side challenges to image and video evidence the prosecution intends to rely on.
- Defence
- Brief
Civil03
Defamation & fraud
Reputation, business-fraud and impersonation matters where synthetic media is the alleged instrument.
- Defamation
- Fraud
Personal04
Image-based abuse
Non-consensual deepfakes and intimate-image abuse — evidence packages for police and eSafety.
- IBSA
- eSafety
The Approach
Confidence,
Confidence,
not certainty theatre.
AI detection is probabilistic. Every Griffin finding is reported with a confidence band and an explicit account of what the model can and cannot tell us about the artefact in question.
Where the matter is serious, a single AI signal is never the case. We combine Griffin output with EXIF, C2PA, recompression history and manual frame review — and say so plainly in the report.
Detection
Deepfake + GAN
Reporting
Confidence-banded
Verification
Multi-signal
Testimony
Court-tested
The Process
Calm, methodical,
Calm, methodical,
court-grade from intake.
Step01
Intake
Original media received under chain-of-custody — provenance, source and prior handling captured.
Step02
Multi-signal analysis
Griffin AI detection combined with EXIF, C2PA and recompression analysis. No single-signal findings.
Step03
Manual review
Frame-by-frame review of video and pixel-level inspection of contested regions by the analyst.
Step04
Report
Plain-English authentication report with confidence bands, limitations and exhibit-ready imagery.
Frequently Asked
Frequently asked about AI authentication.
Request A Consultation
Send A Brief.
Send A Brief.
We'll Take It From There.
Every enquiry is read by a licensed investigator and treated in strict confidence.
Step01
You send a brief
A short note about your matter — no detail required upfront.
Step02
We reply within one business day
From a licensed investigator, not a chatbot or call centre.
Step03
If we're the right fit, we book a call
Confidential. No obligation. Fixed-fee quote where possible.