This AI Application Can Crack Your Password in Less Than One Minute

Passwords are a main aspect of online security, but people often struggle to create strong and memorable passwords. This causes the use of weak passwords that hackers easily compromise. Researchers have developed PassGAN, a machine-learning model that generates strong passwords to address this issue. 

PassGAN is a generative adversarial network (GAN) that uses a training dataset to learn patterns and generate passwords. It consists of two neural networks – a generator and a discriminator. The generator creates new passwords, while the discriminator evaluates whether a password is real or fake.

To train PassGAN, a dataset of real passwords is required. However, using actual passwords presents a security risk. Thus, researchers used a publicly available dataset of password leaks called RockYou, which contains over 32 million passwords that were leaked in a 2009 data breach. The researchers preprocessed the data by removing duplicates, commonly used passwords, and passwords shorter than eight characters. They also added synthetic passwords to the dataset to increase password diversity.

After preprocessing, the dataset was divided into training and testing sets. The training set was used to train the PassGAN model, while the testing set was used to evaluate the model’s performance. PassGAN consists of a generator network and a discriminator network. The generator network inputs a random noise vector and generates a password, while the discriminator network evaluates whether the password is real or fake.

During training, the generator attempts to create passwords that resemble those in the training dataset while the discriminator evaluates the generator’s output and provides feedback on how to improve the generator’s performance. This process continues until the generator can generate indistinguishable passwords from real passwords.

The researchers assessed PassGAN’s performance by comparing the generated passwords to passwords in the testing dataset. They discovered that PassGAN generated passwords that were much stronger than those in the testing dataset.

Although PassGAN can generate strong passwords, it has some limitations. PassGAN is only as secure as the random noise vector used as input. If attackers can predict the noise vector, they can generate passwords that resemble those in the training dataset. PassGAN relies on a dataset of actual passwords to train the model. If the training dataset is compromised, it may result in the creation of passwords similar to the ones in the dataset.

Despite its limitations, PassGAN is a promising approach to generating strong passwords using machine learning. It highlights the potential of machine learning in enhancing online security and serves as a starting point for further research in this field.

Researchers have proposed a few solutions to improve the security of PassGAN-generated passwords. One approach incorporates additional input factors to the generator, such as the user’s age, gender, or occupation. These factors can add randomness and diversity to the generated passwords, making them harder to guess.

Another solution is to use multiple generators, each trained on a different dataset, to generate passwords. This approach can improve the overall strength and diversity of the generated passwords.

Despite these solutions, PassGAN-generated passwords may only be suitable for some applications. For instance, some applications require users to create passwords that are easy to type, which may not be accurate for PassGAN-generated passwords. Therefore, it is essential to consider the application’s specific requirements when choosing a password generation method.

Check out the tool here

Don’t forget to join our 18k+ ML SubRedditDiscord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more. If you have any questions regarding the above article or if we missed anything, feel free to email us at

🚀 Check Out 100’s AI Tools in AI Tools Club

Niharika is a Technical consulting intern at Marktechpost. She is a third year undergraduate, currently pursuing her B.Tech from Indian Institute of Technology(IIT), Kharagpur. She is a highly enthusiastic individual with a keen interest in Machine learning, Data science and AI and an avid reader of the latest developments in these fields.

🚀 [FREE AI WEBINAR] 'Optimise Your Custom Embedding Space: How to find the right embedding model for YOUR data.' (July 18, 2024) [Promoted]