AI Media Authenticity & Deepfake Forensic Workbench

Forensic inspection suite for viral synthetic video and manipulated social media clips

VERDICT: Altered / Deepfake
Video Forensic Frame Viewer
Frame 1 of 3
FRAME: 00:02 | RES: 1080p
NOISE RESIDUAL: 88%
ANOMALY: DETECTED
TIMELINE SCRUBBER 00:02 / 00:08
SENSOR OVERLAY CONTROLS
Detection Telemetry & Provenance
Preset: Deepfake Forensic
COMPOSITE ANOMALY
89.5%
High synthetic deviation
FRAME STATUS
Synthetic Manipulation Detected
2/2 Indicators Flagged
Reported Subject / Context: Steve Sarkisian / Holly Rowe Postgame Exchange
Report Source: @nytimes / @TheAthletic
PRNU Sensor Noise Mismatch: Flagged (Score: 88)
Facial Warping / Boundary Artifacts: Present (Pixel Discontinuity)
Lighting Vector Consistency: Inconsistent (-42 deg delta)
FORENSIC AUDIT BRIEF EXCERPT

Inspection of clip keyframes confirms synthetic temporal discontinuity, boundary blend smoothing around the jawline and hand trajectory, and sensor noise inconsistency across 00:02–00:05. The footage is an AI-manipulated composite.

Reporting Provenance & Verification Context:

Grounding: Texas quarterback Arch Manning publicly addressed an AI-altered viral video depicting Steve Sarkisian and reporter Holly Rowe. Sourced from The New York Times / The Athletic.

Forensic Inspection Pipeline: Decoding AI Media Artifacts

Read the explanation

Digital forensic engines analyze video frames across three critical physical dimensions: sensor photo-response non-uniformity noise, lighting vector alignment, and blending boundary coherence. In altered frames, synthetic models fail physical constraints. The primary stadium keylight and facial cast shadows diverge by forty-two degrees, while boundary blend masks create discontinuous pixel steps. Scrubbing to frame three reveals genuine reference camera noise dropping to forty-two percent, restoring coherent geometry and verifying original broadcast integrity.

Enjoy this tool? Build your own with Super