RTX 5090 cracked most popular passwords in less than an hour - Kaspersky Lab

Almost half of them can be guessed in less than a minute due to the simplest combinations

The new GeForce RTX 5090 graphics card is capable of cracking a huge number of weak passwords in a matter of minutes. Many popular combinations today no longer effectively protect accounts. This was reported by specialists from Kaspersky Lab, who analyzed 231 million unique passwords that leaked into the darknet between 2023 and 2026.

The results were telling: about 48% of passwords were cracked using the RTX 5090 in less than a minute, and 60% in less than an hour. Passwords protected by the MD5 algorithm were used for verification.

The new RTX 5090 showed a brute-force speed of up to 220 gigahashes per second, compared to 164 gigahashes for the RTX 4090. This is approximately 34% faster.

Researchers note that modern graphics cards are making mass brute-forcing of stolen hashes faster and faster, making weak passwords particularly vulnerable. Moreover, attackers don't even need to buy such a graphics card. Computing power can be rented in the cloud for a relatively small amount of money.

The main problem, as researchers point out, is not only the power of the technology, but also the predictability of people themselves. Most weak passwords are short combinations, dictionary words, and simple patterns like "word + numbers + symbol".

The study identified several of the most common patterns:

  • numbers at the end of the password are used by more than half of users;
  • numbers at the beginning are found in almost one in six;
  • years and dates are very often used: every eighth password contains a sequence similar to a year;
  • combinations like 1234 and qwerty are still popular;
  • among special characters, "@" is most common, followed by "." and "!";
  •  passwords often use the words love, angel, team, mate, life, star, and even the meme word Skibidy.

Another conclusion of the study is that users rarely change passwords. According to analysts, 54% of combinations from new leaks have been encountered before. This indicates the reuse of old passwords and identical schemes.

Read more on the topic: