Your analysis will appear here.
The report will include label, score, confidence, evidence, limitations, and recommended next steps.
Gemini-powered anomaly review
Read the signal in synthetic media risk.
Upload an image or video and receive a structured, uncertainty-aware report on classifier risk, visual manipulation indicators, temporal artifacts, and practical next steps.
Upload
Media intake
Files are sent to Google Gemini for analysis. This tool reports risk,
not definitive forensic proof.
Structured output
Risk report
Inspection map
What the analyzer inspects
Classifier signal
Hugging Face fake/real probability for images and sampled video frames.
Boundary artifacts
Face edges, hairlines, ears, glasses, jewelry, and accessory distortions.
Physical consistency
Lighting, shadows, skin texture, teeth, eyes, and facial geometry.
Temporal behavior
Lip sync, blinking, gaze anomalies, identity drift, and flicker.
Quality context
Compression, low resolution, blur, and false-positive risk from poor media.
Important limits
This is risk analysis, not proof.
Treat results as one signal. For high-stakes decisions, combine this output with source verification, provenance checks, metadata review, and expert forensic analysis.
Low-quality media can create false positives.
Authenticity cannot be proven from a single model response.
Gemini receives the uploaded file for analysis.