Agentic AI & Information Integrity

A public research hub connecting open-source artifacts, papers, datasets, and presentation landing pages across synthetic media, provenance, and human vulnerability research.

Program framing

This hub connects a broader research agenda on how agentic and generative AI affect trust, disinformation, verification, provenance, and cognitive resilience on the Web.

The central idea is simple: as synthetic content becomes harder to distinguish from authentic content, trust must increasingly be supported by infrastructure rather than intuition alone.

Core projects

JudgeGPT

Dual-axis human perception platform for evaluating AI-generated news, authenticity judgments, and detection limits.

Human perceptionDisinformationLive demo

Repository · Demo · Zenodo

RogueGPT

Controlled stimulus generation engine for synthetic news and misinformation experiments.

Stimulus generationLLMsResearch tooling

Repository · Zenodo

Origin Lens

Privacy-first mobile provenance verification using C2PA, EXIF, and on-device analysis.

ProvenanceC2PAMobile

Repository · App Store · Preprint

verification-crisis

Expert survey on GenAI disinformation threats and reproducible provenance solutions.

ExpertsGovernanceSurvey

Repository · Preprint · Zenodo

CRED-1

Open multi-signal domain credibility dataset for automated pre-bunking and source-level integrity analysis.

DatasetCredibilityPre-bunking

Repository · Zenodo · Paper

Landing pages

Current preprints and paper links

Datasets and records

Maintained by Alexander Loth. This hub is designed as a QR-friendly entry point for talks, posters, and research dissemination.