Passwords are still the most common form of authentication usedtoday [1]. Most computers and mobile phones available today offerbuilt-in encryption and password protection for storage [2][3]. Passwordcracking plays a significant role in digital forensic investigations.The efficiency of cracking passwords is affected by the password listschosen and can be extremely difficult at times [4]. However, people oftenchoose easy-to-remember passwords that are easy to guess. Usersalso tend to use personal information when creating their passwords,which can be exploited [5]. This study focuses on generating tailoredand customized word lists using PassGAN [6] to improve passwordcracking for Swedish users. We also analyzed patterns, trends, andcontextual information of a collection of Swedish password leaks.Our analysis of 5.8 million Swedish password leaks revealed thatusers tend to opt for simple passwords over long, complex ones,which are often easier to remember. The average password length isapproximately 9.03 characters. When creating passwords, users tendedto either use a combination of letters and numbers or only letters. Thetop passwords were all weak passwords that are easy to guess, suchas "123456" or "hejsan". Swedish-exclusive characters (å, ä ,ö) onlyoccurred around 17,000 times, capitalized or not.Our newly generated password dictionary using PassGAN outperformedrockyou.txt, the well-known password dictionary. Merginglocal elements such as names and expressions with real-life passwordleaks contributed to enhancing the model’s performance in predictingand generating realistic passwords.