Forensic-grade  ·  Published per-use-case FPR  ·  Audit-grade chain-of-custody  ·  Abstains when unsure

Deepfake detection platform.
Built for what's at stake.

Reproducible, explainable, and traceable analysis across face manipulation, AI-generated content, and synthetic voice — backed by reproducible, version-stamped chain-of-custody bundles and a per-use-case error rate we publish.

Face Manipulations
Synthetic Voices
AI Generated Videos
AI Generated Images
2
Live dimensions
5 on roadmap
99.75%
Image recall
@ 0.5% FPR
90s
Average full
analysis time
4
Heatmap variants
version-stamped
Submit for analysis
Drop file here to analyse
Video, image · up to 150 MB · analysed and deleted within minutes
mp4movwebmpngjpg
Recent submissions
🎥
interview_video_nov_14.mp4
#67201 · 2h ago
Suspicious
🎥
ceo_call_recording.mov
#67200 · 5h ago
High risk
📸
board_call_jan.mov
#67198 · 2d ago
Authentic
Face manipulation · 4 attack categories

Every face attack.
Exposed and explained.

Identifies gen-AI face manipulation by analysing visual artefacts, biometric inconsistencies, and generative patterns. Four attack categories. Minimum three sub-signals per finding. Every result linked to a plain-language reason code.

Boundary artefact score
0.88
Blending seam detection
0.82
Identity embedding dist.
0.75
Spatial frequency mismatch
0.69
Lip-sync alignment
0.80
AI face swapAI lip syncFace reenactmentFace morphing
Face analysis · BWV-HM-FACE v2.0SUSPICIOUS
Boundary anomaly
Seam artefact
engine v2.0 · heatmap v2.0