# Seeing is Not Always Believing: An Empirical Analysis of Fake Evidence Generators

**Link:** <https://ieeexplore.ieee.org/document/10628523>

**Conference:** EuroS\&P 2024

**Keywords:** Fake Evidence, Deepfake, Image Forensics

## Summary

Build detector and conduct a empirical study on the ecosystem of fake evidence generators (FEGENs).

## Threat Model

FEGEN

![alt text](https://169318583-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUAYVuGJIcqGBOPcOsA14%2Fuploads%2Fgit-blob-cc8f54f82fe9e5de1349c7cc33ba384aece503ee%2Fimage-4.png?alt=media)

![alt text](https://169318583-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUAYVuGJIcqGBOPcOsA14%2Fuploads%2Fgit-blob-d726e2df5c3824805bb2b3e20f2b10e7aadbee01%2Fimage-5.png?alt=media)

## Data

1. Review public reports to get categories of potential FEGENs ![alt text](https://169318583-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FUAYVuGJIcqGBOPcOsA14%2Fuploads%2Fgit-blob-a96698605768c65d6883124886a9c407766d468d%2Fimage-6.png?alt=media)
2. Search on google, facebook, and tiktok. Get 148 instances with 72 AAS websites, and 76 software tools, corresponding to the 125 potential FEGENs

## Qualitative + Quantitative Analysis + Case study

(Note: Telegram search: SuperIndex News, Telegram Channels, TGStat)