Skip to main content
  1. Home
  2. Computing
  3. News

AI can probably crack your password in seconds

Add as a preferred source on Google

We can now add easily cracking passwords in a matter of seconds to the list of things that AI can do.

Cybersecurity firm Home Security Heroes recently published a study uncovering how AI tools analyze passwords and then use that data to crack the most common passwords used on the web.

Recommended Videos

Using the PassGAN tool, the firm was able to figure out common four- to seven-character passwords in seconds. It also didn’t matter if there was variation in uppercase and lowercase letters or if numbers were included. The shorter and more simple passwords were easier for the tool to crack.

PassGAN uses the latest Generative Adversarial Network (GAN) machine learning model that has been fed over 15 million common passwords. These passwords have been derived from the RockYou data set, which has collected information from popular breaches of companies such as MySpace and Facebook. The RockYou data set has become a commonplace source for machine learning password-cracking models, according to Tom’s Hardware.

PassGAN was able to crack passwords with up to six characters instantly until symbols were included — for those, it took at least four seconds. The tool was able to crack passwords with up to seven characters instantly until they included uppercase and lowercase letters; then it took at least 22 seconds.

Overall, the study determined that passwords longer than 12 characters with a combination of uppercase and lowercase letters, numbers, and symbols were the most challenging to crack. For example, a 15-character password with such a mix would take 14 billion years for AI to crack, according to PassGAN.

However, in common practice, most users are still very much at risk for a password breach. Home Security Heroes notes that for most of the common passwords, at least 51% of those tested were cracked in less than a minute. Many that are more challenging can still be figured out with time; 65% of common passwords can be cracked in less than an hour, 71% in less than a day, and 81% in less than a month.

Tom’s Hardware noted via Statista that six out of 10 Americans have a password between eight to 11 characters. While an 11-character password with uppercase and lowercase letters, numbers, and symbols can put you in the safe zone of 356 years to crack, many users might still be at risk with shorter, less unique passwords.

Users should keep in mind common password safety practices such as not keeping the same passwords for multiple accounts, changing passwords regularly, and using trusted password managers.

Fionna Agomuoh
Fionna Agomuoh is a Computing Writer at Digital Trends. She covers a range of topics in the computing space, including…
Apple’s M6 chip isn’t even here yet, but you’ll see M7 Macs early in 2027
Apple is reportedly already accelerating its next-generation silicon roadmap, even before the M6 has launched.
Apple MacBook

The M6 chip is still expected to debut later this year, but Apple may already be preparing for what comes next. According to Mark Gurman's latest report for Bloomberg, the company is aiming to introduce its first M7-powered devices as early as the first half of 2027, hinting at a much faster silicon refresh than many expected.

M7 could arrive alongside new Macs and iPads

Read more
The entry-level MacBook Pro could get a design refresh in 2027, and it’s about time
Five years on the same chassis, and now both tiers of the MacBook Pro are getting a new look at once.
MacBook Pro in space grey sitting on a desk.

Apple has a new MacBook Pro lined up for launch early next year, according to Bloomberg. The company will introduce a 14-inch laptop in the first half of 2027. 

The biggest surprise, however, will be a brand-new design language. The outlet describes it as "a revamped entry-level MacBook Pro, code-named K104."

Read more
Study finds humans will talk to AI ghosts of the dead as reincarnations, and it’s pretty grim
The first AI ghost study is in. The results are about as complicated as you'd expect.
VR Headset, Person, Face

A new study from the University of Colorado Boulder confirms something that sounds both impressive and concerning. People find interacting with AI simulations of their dead loved ones deeply meaningful, and most will come away wanting to do it again.

The researchers call it a "generative ghost," which is a clear reference to generative AI, but I’d still prefer to call it unsettling.

Read more