• Version
• Download 2
• File Size 438.29 KB
• Download

Authenticity Lens: An AI Media Detection System for Verifying Content Authenticity

   T. Amalraj Victoire 1, A. Jagathrachagan 2 , N. Harish 3

1Associate Professor, Department of Computer Applications, Sri Manakula Vinayagar Engineering College (Autonomous), Puducherry 605008, India,

amalrajvictoire@gmail.com

2Post Graduate student, Department of Computer Applications, Sri Manakula Vinayagar Engineering College

(Autonomous), Puducherry 605008, India,

jagathrachagan285@gmail.com

3Post Graduate student, Department of Computer Applications, Sri Manakula Vinayagar Engineering College

(Autonomous), Puducherry 605008, India,­

  harihz2001 @gmail.com

Abstract

In the age of advanced artificial intelligence (AI), the proliferation of AI-generated media has raised concerns regarding content authenticity. With AI tools capable of creating realistic images, videos, audio, and text, it has become increasingly difficult to distinguish between genuine and fabricated media. This challenge has led to the development of the "Authenticity Lens: AI Media Detector", a comprehensive solution designed to identify AI-generated content across multiple media formats and ensure the integrity of digital information.

The project consists of four main modules: AI Image Detection, AI Text Content Detection, AI Audio Detection, and Content Integrity Verification for Images. The AI Image Detection module utilizes a pre-trained MobileNetV2 model to classify images as either AI-generated or real. By analyzing visual features, it accurately detects AI-manipulated images that may deceive viewers. The AI Text Content Detection module employs tools like spacy and Text Blob to examine the structure, tone, and fluency of social media posts or other user-generated text. It identifies repetitive phrases, unnatural writing patterns, and other indicators of AI-generated text, helping to combat misinformation in online environments.

The AI Audio Detection module, based on the YAM Net model, classifies audio files as either AI-generated or real by analyzing audio features. This helps in detecting deepfake audio or synthetic voices. Additionally, the Content Integrity Verification module checks images for signs of tampering, using techniques like perceptual hashing to identify image manipulations that may indicate fake or altered content.­­

Built on a Flask-based web application, the system allows users to upload images, text, and audio files for analysis. The backend leverages pre-trained models, such as MobileNetV2 for image classification and YAM Net for audio classification, along with spacy and Text Blob for text analysis. The project provides an efficient, user-friendly solution for detecting and verifying AI-generated content in real-time, offering an essential tool in the fight against digital misinformation.

Although the core modules are complete and operational, future work will focus on improving model accuracy, expanding the dataset for more robust detection, and scaling the system to handle larger volumes of data. With ongoing advancements in AI technology, Authenticity Lens offers a proactive approach to preserving media authenticity and ensuring trust in digital content.

Key words: AI-generated media, content authenticity, AI tools, digital information, Authenticity Lens, AI Image Detection, AI Text Content Detection, AI Audio Detection, Content Integrity Verification, MobileNetV2, image classification, spacy, Text Blob, social media analysis, misinformation, YAM Net, audio classification.