home.social

#passgan — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #passgan, aggregated by home.social.

  1. Terrifying study shows how fast #AI can #crack your #passwords
    51% of all common passwords were cracked in less than one minute, 65% in less than an hour, 71% in less than a day, and 81% in less than a month. They used new #password #cracker #PassGAN (password generative adversarial network) to process over 15M credentials from the Rockyou dataset. Only passwords with 18 characters that include both letters and numbers were found to be safe from AI cracking for now.
    9to5mac.com/2023/04/07/ai-crac

  2. @trusttrist We've been able to crack 7 character passwords that quickly for a LONG time, and much faster than #PassGAN, well before PassGAN was a thing. The math is very straightforward, and PassGAN is actually a step backwards. Here's an example of the math: grc.com/haystack.htm

  3. isn't really doing anything new we haven't already been doing with Markov chains when it comes to password cracking. Don't buy into the hype.

    arstechnica.com/information-te

  4. #AI based #password candidate generation is super interesting to me, but so far the best generators (CMU's #neuralnetwork based generator presented at #USENIX 16, and Dr Matt Weir's #PCFG generator) are still about on par with existing #Markov implementations and rule-based #cracking.

    #PassGAN, on the other hand, falls well short of existing password cracking techniques and candidate generators.

    While I hold out a lot of hope for future AI-powered password cracking, it ain't here yet, and it definitely isn't PassGAN.

    arstechnica.com/information-te