Detecting AI-Generated Personas

Read full text Summarise with VIRAT
Authors: Karen Guzman
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This repository contains the full pipeline for detecting AI-generated social media personas. The project combines two datasets: Instagram account metadata from the InstaFake dataset and face images from the Kaggle Human Faces Dataset. The `data` folder contains `fakeAccountData.json` and `realAccountData.json` from the InstaFake dataset. The data is normalized into 4 relational tables (accounts, account_stats, account_profile, face_images), stored in Parquet format within the `data` folder, and loaded into DuckDB for querying. A fine-tuned MobileNetV2 CNN classifies profile pictures as real or AI-generated, and a Random Forest classifier detects fake accounts from the metadata. Predictions from both models are combined into a final detection score. Additionally, the license and press release are displayed in the root repository, alongside the jupyter notebook that holds the code for the pipeline labeled `Pipeline.ipynb`.

Fields of Study & Subjects

Digital Media Forensic Detection Spam and Phishing Detection Misinformation and Its Impacts

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
31 March 2026
Language
EN