Abstract
Respondus LockDown Browser (RLB) is a widely recognized tool employed to enhance the security of online examinations by thoughtfully restricting certain system functionalities. Despite its effectiveness, recent research has uncovered IP-based Keyboard-Video-Mouse (IPKVM) exploits that enable remote cheating through hardware-level interventions undetectable by current proctoring software. This paper presents a machine learning (ML)-based detection method analyzing mouse movement patterns and input behavior to identify remote-controlled cheating. The prototype model achieved 95% accuracy in detecting remote control attempts, establishing a foundation for behavioral biometrics in exam security with applications extending to cybersecurity. This approach enhances exam integrity while minimizing privacy concerns and advancing secure and fair assessment in digital testing environments.