Document Detector
PDF, DOCX and TXT — content and provenance together
Documents carry more than words. TraceMark AI reads the writing patterns and the document's own metadata, and reports each separately.
Drag and drop a file to analyze
or click to browse — PDF, DOCX, TXT, JPG/JPEG, PNG
Upload progress is simulated locally. No file leaves your device and no analysis runs in this build.
Writing-pattern analysis
- Token surprise (perplexity)Pending engine
Measures how predictable each word is given the preceding context.
- Sentence-length variance (burstiness)Pending engine
Human writing typically varies more in sentence rhythm.
- Lexical diversityPending engine
Ratio of unique to total vocabulary across the passage.
- Connective densityPending engine
Frequency of formulaic transitions such as 'moreover' and 'therefore'.
Metadata & provenance analysis
- Document producerPending engine
Authoring tool recorded in the file metadata.
- Edit history depthPending engine
Revision trail found inside DOCX/PDF structures.
- C2PA / content credentialsPending engine
Signed provenance manifests when present.
- Image generation markersPending engine
Resampling and noise-fingerprint checks for JPG/PNG.
AI detection is probabilistic. This result does not establish authorship and should not be used as the sole basis for academic, employment, disciplinary, or legal decisions.
Handling
Built for files that matter
Documents submitted for review are often sensitive. Retention, access and export are all controlled by your organization.
Configurable retention
Delete originals immediately after analysis, or keep them for a fixed review window.
Scoped access
Only reviewers and administrators in the owning workspace can open a submission.
Exportable record
Reports capture the evidence, the reviewer note and the case status at export time.