Home » Technology » AI Security Threat: How Your Passwords Can Be Stolen with 95% Accuracy

AI Security Threat: How Your Passwords Can Be Stolen with 95% Accuracy

Jakarta, CNBC Indonesia The research team from Cornell University discovered a serious security threat from the application of artificial intelligence (AI) technology. According to studies, AI tools are able to steal people’s important passwords with an accuracy of up to 95%.

The theft was carried out simply by hearing the sound of a laptop keyboard typing. To test this, researchers trained an AI model to learn the sound of keyboard typing on the MacBook Pro.

The typing sound enters through the cellphone microphone, then the data is collected to build an AI detection algorithm, quoted from DigitalTrends, Tuesday (8/8/2023).

ADVERTISEMENT

SCROLL TO RESUME CONTENT

The research team also tested its accuracy while making Zoom calls. The sound of keyboard typing is recorded using a microphone on a laptop during the meeting.

AI is capable of producing up to 93% accuracy in reproducing keyboard sounds to identify passwords. On Skype, the accuracy can be up to 91.7%.

It should be noted, the keyboard typing sound is not identified from the volume set by the user. The accuracy is learned from the sound of the typing waves, the intensity, and the timing of the beats, which are matched to the typing habits of the majority of people.

So, there’s no point turning off the mechanical keyboard audio. When this AI tool is inserted on a device that has a microphone, cybercriminals can use it to find out your important passwords.

Last week, the FBI just gave a warning about the bad impact of AI tools, including ChatGPT, for the security of internet users. Cybersecurity experts also say tools like ChatGPT can create more adaptive malware.

[Gambas:Video CNBC]

Next Article

bad! ChatGPT Admits Leaked User Credit Card Data

(fab/fab)

2023-08-08 12:30:00
#Study #Reveals #Steals #Password #Instantly #Watch

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.