Research directions
Synthetic Media Forensics
Examination of generated or altered image, video, audio, and multimodal content.
Read referenceReferenceAI-Generated Artifacts
Generated files and residual application, device, and agent records surrounding AI use.
Read referenceReferenceProvenance & Authenticity
Origin, lineage, transformations, and the support for claims about what an object is.
Read referenceReferenceDetection & Attribution
Testing synthetic-content signals and associations with a source, model, workflow, or actor.
Read referenceReferenceEvidence Reconstruction
Recovering supported event sequences and source relationships from incomplete digital records.
Read referenceReferenceHuman-AI Interaction
Records and interpretation of exchanges between people and AI-enabled systems.
Read referenceReferenceSynthetic Memory & Source Monitoring
Research on how people distinguish experienced, communicated, imagined, and synthetic information.
Read referenceReferenceChain of Custody & Evidence Integrity
Preserving bytes, handling history, versions, and traceability of original and derived evidence.
Read referenceReferenceAI Anti-Forensics
Defensive study of interference with the availability or interpretation of AI-related evidence.
Read referenceReferenceLegal & Evidentiary Issues
Authentication, method reliability, disclosure, and the limits of forensic interpretation.
Read referenceResearch questions
- Model-family attribution with unknown-source rejection and realistic transformations.
- Synthetic video lineage and provenance survivability after redistribution.
- Timeline reconstruction across asynchronous tools, retries, and delegated agents.
- Source-memory effects and authenticity calibration under ethical controlled designs.
- Evidence contamination, custody of agent-generated artifacts, and reproducible derivations.
- Confidence calibration, cross-model transfer, and transparent negative results.
Platform workflows
Accounts support verified access, mandatory second-factor authentication, scoped roles, and private profile defaults. Tool and course listings use human approval. Native learning retains enrollment and progress, while credential records publish their actual criteria and status. Public search draws from explicitly public reference and released content.
Publishing and resource discovery
Blog and Journal authors work in private drafts and submit exact versions for assigned human review. Approved releases retain immutable publication records and correction history. Topic pages connect the reference with related public resources, and the resource directory links tools, courses, datasets, and publications by purpose.
Help with a bounded contribution
Choose a research area, identify the question and required evidence, and explain a feasible test. The project inventory and preparation guide connect available resources with a clear contribution.
Canonical reference: https://syntheticforensics.org/roadmap/
Report a technical correction