AI has reached a new level of sophistication in its ability to steal passwords, according to a recent study. Researchers have found that an AI tool can decipher text, including passwords, from keystroke sounds recorded over Zoom with an astonishing accuracy rate of 93%. The rate rises even higher, up to 95%, when the keystrokes are recorded using an iPhone 13 mini. This alarming discovery highlights the growing threat of “acoustic side-channel attacks” on keyboards, where hackers exploit information such as keystroke sounds rather than directly attacking the system’s code. As AI tools become more advanced, concerns over security and privacy are mounting, raising questions about the vulnerabilities of internet-connected devices.
AI Can Steal Passwords from Keystroke Sounds Recorded Over Zoom with High Accuracy
A new study conducted by researchers from Durham University, the University of Surrey, and the Royal Holloway University of London has found that an AI tool can steal passwords from keystroke sounds recorded over Zoom with up to 93% accuracy. The accuracy rate even rose to 95% when keystrokes were recorded using an iPhone 13 mini.
The Threat of Acoustic Side-Channel Attacks
The researchers highlighted the growing threat of “acoustic side-channel attacks” to keyboards. These attacks exploit information such as keystroke sounds to gain unauthorized access to sensitive data, rather than directly attacking the system’s code. The AI model developed by the researchers, based on deep learning, was able to decipher keystrokes from a recording of a MacBook’s keystrokes made over Zoom with an accuracy rate of 93%.
Potential Countermeasures and Vulnerabilities
The study also discussed potential countermeasures to mitigate the risk of acoustic side-channel attacks. The researchers found that passwords containing full words are at a greater risk of being compromised. They also discovered that touch typing and adding background noise can lower the accuracy rate of the AI tool.
Interestingly, the study noted that these types of attacks have a long history, with “acoustic emanations” being mentioned as a vulnerability in a 1982 partially declassified NSA document. This highlights the need for continued research and awareness of such threats.
AI Tools and Security Concerns
The findings of this study contribute to the growing concerns about how AI tools can be used to compromise security and privacy. AI makes it easier for online scams to be personalized for each target, making them harder to detect. Additionally, specialists have previously raised concerns about how the advancement of AI and 5G technology can heighten vulnerabilities in internet-connected devices, amplifying cybersecurity threats.
In conclusion, this study highlights the potential risks associated with acoustic side-channel attacks and the need for stronger security measures to protect sensitive information. Users should consider using strong, unique passwords that do not contain full words, and organizations should invest in technologies that can detect and prevent these types of attacks. As technology continues to advance, it is crucial to stay vigilant and proactive in safeguarding our data and privacy.
Takeaways:
- Acoustic side-channel attacks pose a growing threat to keyboard security.
- An AI tool can decipher keystrokes recorded over Zoom with up to 93% accuracy.
- Countermeasures such as avoiding passwords with full words and adding background noise can lower the risk of attacks.
- AI tools can make online scams more personalized and harder to detect.
- Continued research and awareness are necessary to address the vulnerabilities posed by AI technology.
