Unite.AI

Easy Rewording Breaks AI Safety, Even for Gemini and Claude

AI safety tests found to rely on 'obvious' trigger words; with easy rephrasing, models labeled 'reasonably safe' suddenly fail, with attacks succeeding up to 98% of the time. New corporate research ...
Cyber Defense Magazine

Deepfakes, Synthetic Media, and Digital Trust: The Cybersecurity Implications of Deepfake Technology and Methods for Detection and Mitigation

Introduction The proliferation of deepfake technology, synthetic media generated using advanced artificial intelligence techniques, has emerged as a ...
IEEE

Adversarial Machine Learning for Robust and Secure UAV Detection in Consumer Applications

Abstract: This research evaluates a cognitive AI model for unmanned aerial vehicles (UAV) detection using adversarial machine learning (AML) techniques. We test the model using the VisDrone dataset ...
IEEE

Detecting and Mitigating Adversarial Machine Learning Attacks in Autonomous Vehicles Within the Internet of Vehicles

Abstract: Adversarial Machine Learning (AML), particularly model poisoning, presents a critical threat to Autonomous Vehicles (AVs) in the Internet of Vehicles (IoV) environment. To address this ...