AI Capable of Identifying Typed Passwords Through Sound Analysis
Shared By Peter Boykin – American Political Commentator / Citizen Journalist
AI Capable of Identifying Typed Passwords Through Sound Analysis
AI Capable of Identifying Typed Passwords Through Sound Analysis
Recent research from Cornell University has demonstrated that artificial intelligence (AI) can achieve nearly 100 percent accuracy in identifying computer passwords by analyzing the sounds produced when keyboard keys are pressed. The study involved training an AI computer system to recognize the distinct audio patterns generated by MacBook Pro keys during typing activities over both phone calls and Zoom conference calls.
The AI system rapidly acquired the ability to differentiate between the unique sounds of individual keys. It achieved a remarkable 95 percent accuracy in sound identification during phone calls and 93 percent during Zoom calls. This development highlights the potential vulnerability that could be exploited for cyberattacks utilizing this technology.
Dr. Ehsan Toreini, the co-author of the study, acknowledged the increasing accuracy of such AI models, suggesting that the risk of potential cyber threats involving sound-based password detection is likely to grow over time.
Professor Feng Hao from Warwick University echoed this concern, emphasizing the significance of both auditory and visual cues during virtual interactions. He cautioned against typing sensitive information, including passwords, on keyboards while engaged in Zoom calls. Professor Hao further highlighted the potential of visual cues, such as subtle movements of the shoulder and wrist, to inadvertently disclose information about keystrokes, even when the keyboard itself is not visible on camera.
The findings of this research underscore the evolving landscape of cybersecurity challenges, urging individuals to adopt heightened precautions to safeguard sensitive information in virtual environments.
[Source: Cornell University Study, The National Pulse]
