[ad_1]
You may think that hiding your screen while typing passwords is enough to keep your data safe. But new research shows that even the sound of your keystrokes can be used to steal your personal information. Artificial intelligence (AI) can now crack passwords with over 90% accuracy simply by listening to someone type.
Hackers use AI to interpret keystroke sounds
The concept of using audio recordings to discover passwords is called an ‘acoustic side channel attack’ (ASCA). It was first studied decades ago, but was not taken very seriously. However, advances in AI and the rise of video calling now make this a legitimate threat.
ASCAs allow hackers to collect information externally rather than having to infiltrate a device directly. Interpreting sounds as keyboard clicks can gather enough data to decipher what is being typed.
The research that shows dangerous levels of accuracy
A recent study tested AI’s terrifying ability to crack passwords via acoustic side-channel attacks. Researchers used a smartphone to record keystroke sounds from a MacBook Pro. Shockingly, the AI model was able to reproduce passwords with 95% accuracy just by listening.
Even during Zoom calls, the AI achieved 93% accuracy by listening through the laptop’s microphone. This shows that videoconferencing makes people even more vulnerable to these attacks.
ALSO READ: Is Google Gemini AI available in India?
More than just passwords at risk
Although the study focused specifically on passwords, experts say acoustic side-channel attacks can also be used to steal:
- Credit card numbers
- Bank account information
- Sensitive documents
- Private messages/emails
- Any other typed text
So how does it work? The AI listens to the unique sound of each keystroke and analyzes details such as:
- Which keys produce higher/lower notes?
- Difference in force/pressure between keystrokes
- Speed and rhythm of typing
- Finger used for each keystroke
With enough data, the algorithms can match sound patterns with specific keys and words with high accuracy.
Increasing access to audio recording devices
What makes this threat especially concerning is the ubiquity of microphones in modern devices. Here are just a few everyday items that could potentially be recording your keystroke sounds without you realizing it:
- Smartphones
- Smart speakers/home assistants
- Laptops
- Security cameras
- Tablets/iPads
- Gaming consoles
It’s simply impossible to avoid having microphones nearby when you live in the modern, connected world. Even if you mute yourself during calls, you won’t be fully protected if a hacker has compromised nearby devices.
Tips to protect yourself
While completely preventing audio recording is unrealistic, there are some steps users can take to minimize the risk of acoustic side-channel attacks:
1. Add background noise
Turning on a fan, radio, or white noise machine near your workspace makes it harder for AI models to isolate and interpret keystroke sounds.
2. Use noise-cancelling headphones
Wearing headphones with active noise cancellation adds an extra layer of acoustic protection when typing sensitive information.
3. Choose biometric authentication
Using fingerprint or facial recognition logins instead of typed passwords removes the audio component entirely.
4. Change passwords regularly
Regularly updating passwords ensures that any compromised credentials have a short lifespan.
READ ALSO: Can I use Gemini AI now?
The need for increased security measures
This study highlights the cold reality that our personal data is becoming increasingly vulnerable to AI-driven cybercrime. Typing something as mundane as a username suddenly carries major risks in the age of smart devices and acoustic side-channel attacks.
Just as antivirus software has evolved to combat malware, new protective measures must now be developed aimed at detecting and preventing audio recordings. Policymakers must also address the legal gray areas surrounding the use of logging devices, which currently give hackers the opportunity to exploit them.
User awareness about threats such as ASCAs is the first step. But tech companies and authorities should also consider acoustic surveillance as a growing public concern. Failure to implement safeguards against AI/audio-based hacking endangers the personal, financial and psychological well-being of millions of people in today’s digital society.
🌟 Do you have burning questions about “Can AI crack passwords by listening”? Do you need some extra help with AI tools or something else?
💡 Feel free to send an email to Arva, our expert at OpenAIMaster. Send your questions to support@openaimaster.com and Arva will be happy to help you!