Research led by Cornell University and other prestigious institutions has made light of a new AI-powered method that can identify passwords by simply listening to typing on a keyboard.
This study demonstrated the potential risks posed by everyday devices as well as the application of artificial intelligence in cybersecurity.
How Does This Work?
The method uses AI to study the unique sounds each keyboard key makes. By identifying these sounds, the AI can guess what’s being typed.
The core of the experiment relied heavily on the advanced microphones found in modern smartphones and laptops.
These microphones, with their enhanced sensitivity, could detect and differentiate between the sounds produced by keys on a MacBook Pro.
The Accuracy of AI
The result was a success rate of 95% in identifying the exact keystrokes, showing how accurate this AI-driven method is.
More tests were conducted over popular video communication platforms, Zoom and Skype. The results were still impressively high, as the AI achieved 93% accuracy on Zoom calls and about 92% on Skype.
These figures highlight the technique’s effectiveness, even with potential background noises and other audio distractions present during such calls.
More from News
- I Woke Up To A Meta Account. When Did I Agree To That?
- Vidoser Launches AI-Powered Platform To Streamline Creator Marketing For Brands And Agencies
- Office Freedom Named Agency Of The Year And Deal Of The Year 2026
- A “Power Event” At Google Cloud’s London Region Affected Revolut And European Banks – What Happened?
- Anchor Launches Action Index, Cutting Computer-Use Agent Token Costs By 100X
- Anthropic Selects Build Circle As A UK Partner To Help Supercharge AI Realisation For Clients
- An OpenAI Engineer Is Charged Over A $14 Recording Device – What Does It Say About Security At AI Labs?
- Mhance Acquires CPS And City Dynamics To Strengthen Its Copilot, Modern Work And Business Solutions Capabilities
The Challenges that Prevent the AI’s Efficiency
The AI method, while showing promising results, isn’t perfect. One major inefficient step is the need for the AI to train separately for each type of keyboard, meaning it can’t be broadly applied to every keyboard out there.
This will also meam, though, that once trained, the volume or brand of a keyboard wont provide protection either. Even the models that have silent keyboards are vulnerable to this advanced listening technique.
Securing Against Advanced Threats
Shifting to biometric security options, like fingerprint or facial recognition, can offer stronger defense against such cyber threats. Modern technology gives users alternative security tools.
Features like Windows Hello and Apple’s Touch ID offer both convenience and security, reducing the need for typing out passwords.
Trusted password managers can also be useful, because users have the ability to store their information securely. These tools also usually have options to suggest diverse, strong passwords for you.
These findings are a reminder for all internet users to protect their data and passwords as we are witnessing cybersecurity become more and more vulnerable.
Cyber criminals are more capable than ever to compromise just anybody’s privacy, but equally, many resources and precautionary tools are made for us to prevent breaches.
Minimising Keyboard Detectors Accuracy
According to research, one of the most effective strategies to prevent these AI-driven keystroke recognition attacks is to change one’s typing style.
Interestingly, touch typing, which relies on muscle memory rather than sight, was found to lower the AI’s accuracy in identifying keystrokes, dropping it from 64% to 40%.
Experts also recommend using randomised passwords with a variety of cases, symbols and numbers, to better prevent the risk of potential cyber threats.
