Twitxy Saved

mRr3b00t

Verified account This account is verified. Learn more

@UK_Daniel_Card

Chief Artificial Intelligence Cyber Security Scientist

Superintelligence-grade security

former Helpdesk

AI infection status: clean

Task Force 29
London, Vauxhall pwndefend.com Joined April 2009
8.5K Following 127K Followers 495K Posts

Latest posts 5

mRr3b00t Verified account @UK_Daniel_Card
I've been testing LLMs and using them to help with security tasks etc. for over a year now..... I'm not even sure how long....

Everything I'm seeing here broadly is:

they are really good CO-PILOTS

they aren't so great for just letting them run off and attack shit....

because they create a mess. they are noisy, they get caught, machines get isolated (if you have a decent posture)

I'm struggling to be amazed...... impressed......

I dunno gang, the hype people give vs what I see on actual systems (and from incident intelligence)...

if you run these against a system with shit posture they will TRASH it... sure! congrats!