hi, i'm not rly here much. sorry to say but i'm much more active over on bluesky: https://bsky.app/profile/mussar.io
i'll try to be more active and will definitely bridge that account if i haven't already, but expect me over there
| BirdApp | https://twitter.com/MussarMusic |
| GitHub | https://github.com/mussar0x4D5352 |
hi, i'm not rly here much. sorry to say but i'm much more active over on bluesky: https://bsky.app/profile/mussar.io
i'll try to be more active and will definitely bridge that account if i haven't already, but expect me over there
@daringfireball macOS comes with Tetris pre-installed since Catalina. It's built into zshell:
autoload -Uz tetriscurses
tetriscurses
🦀 Have you been noticing that over the past two years, there seems to be more and more malware written in Rust? Have you ever wished there was one page that collected all these malware families and samples in one place, so that you could practice your Rust reverse engineering?
Here is that place! The Rust Malware Sample Gallery: https://github.com/cxiao/rust-malware-gallery
I have collected information about every Rust malware family that I could find, and scoured public malware repositories to find at least one public sample that is available for each family. Download links are provided for each sample to MalShare or Malware Bazaar, neither of which require an account for you to download the samples.
I hope that this can help reversers get a better understanding of Rust binaries, and improve the state of the art in Rust malware reversing. We have a long way to go 😅
Please send a pull request if you notice something is incorrect or missing! Happy reversing!
#MalwareAnalysis #malware #ReverseEngineering #ThreatIntel #cybersecurity #rust #rustlang #infosec
@joshbressers @kurtseifried @norabatem there's a broader conversation here about the potential of machine learning technology as a whole, in that it's an application of probability/statistics.
There is in fact a limit to accuracy in all ML models, and that fuzziness is both an asset and a detriment. Most of the advancements in digital photography come from the fact that ML models are the best kind of dithering we've found. The fuzziness of LLMs are what makes them capable of more than just regurgitating their training data word for word.
We want that inaccuracy, we just also want to slap it on things that require more accuracy than it can provide, and when that accuracy exceeds what we could do ourselves we risk falling into the trap of believing that it's "correct".
@joshbressers @norabatem @kurtseifried yeah, sorry for not responding sooner - phone was on "spend time with significant other" mode soon after sending that.
i don't mind the usage of gen ai images but you just pointed out the biggest flaw in using them - they only look good if you don't care enough about visuals to look at it for long enough to notice all the flaws.
same's true for all generative machine learning models, it's just more pronounced in images
Remember! To make your Mickey Mouse original content legal, he needs to be the iteration from Plane Crazy and Steamboat Willie.
Distinctions:
Smaller ears than modern mickey
No gloves
Work boots instead of potato shoes or, alternatively
Plain black feet
Optional (but helpful for copyright!) boat captain hat
Slightly elongated pill-shaped nose
Long tail
(Optional) buggy eyes
White pants! Or other colored pants but NOT red or you're treading into modern mouse territory