Adam R

Musings on Life, The Universe and Everything. Principal Consultant and CISO, CyberSec. Runner, science nerd. Reading Phil. of Science. See:

[Adam R @reason42]

@RobKenyonReform @antmiddleton Green screen video.

Also Middleton is a convicted criminal

In 2013, he was convicted of:
• Unlawful wounding (of a male police officer).
• Common assault (of a female police officer).

So, yeah…. appropriate endorsement.

@supertanskiii ☝️

[Adam R @reason42]

@UK_Daniel_Card Bruh. Same problem. I get restless when I have loads of work ideas / problems to resolve.

[Adam R @reason42]

I like this idea... ...might put such content in white text on my LinkedIn Profile @infosec_fox 🤣

[John Scott-Railton @jsrailton]

NEW: malware developers added nuclear & biological weapons text to to their spyware.

Goal? To trigger LLM safety refusals... so that their spyware wouldn't be analyzed by an AI security scanner.

Cleanest practical example I can think of for why over-indexing on first order safety alignment is risky.

When closed (and open) models ship with aggressive refusals, they will be sprinkled with second-order blindspots that attackers will discover...and exploit.

We are only in the earliest days of attackers leveraging these features, and it wouldn't surprise me if users systems that need to handle complex cybersecurity issues demand that models be less safety-blunted.

In the weeds: @SocketSecurity's post also shows why intention matters in how you design a malware analysis pipeline to avoid prompt manipulation.

H/T to colleagues that shared this with me https://t.co/f3Aj9TYxU4

[John Scott-Railton @jsrailton]

NEW: malware developers added nuclear & biological weapons text to to their spyware.

Goal? To trigger LLM safety refusals... so that their spyware wouldn't be analyzed by an AI security scanner.

Cleanest practical example I can think of for why over-indexing on first order safety alignment is risky.

When closed (and open) models ship with aggressive refusals, they will be sprinkled with second-order blindspots that attackers will discover...and exploit.

We are only in the earliest days of attackers leveraging these features, and it wouldn't surprise me if users systems that need to handle complex cybersecurity issues demand that models be less safety-blunted.

In the weeds: @SocketSecurity's post also shows why intention matters in how you design a malware analysis pipeline to avoid prompt manipulation.

H/T to colleagues that shared this with me https://t.co/f3Aj9TYxU4

[Adam R @reason42]

..into the rabbit hole I go....

[Adam R @reason42]

@UK_Daniel_Card lol, I just left London. Back home in the Shires now. I've left behind the sunshine for you. You're welcome.

[Adam R @reason42]

@supertanskiii Seems to me they’re both trying to out grift each other by lurching further and further to the hard right… exploiting the low-information public for their greed.

[Adam R @reason42]

@LisaForteUK Noted 😊. Are you ever at any GTIA events?

[Adam R @reason42]

@supertanskiii Seems to me they’re both trying to out grift each other by lurching further to the hard right… exploiting the low-information public to their greedy grips.

[Adam R @reason42]

@davemccollough @UK_Daniel_Card There’s beauty in simplicity. Nice ride! And also trail 🏆

[Adam R @reason42]

@UK_Daniel_Card The guy was triggered because I went in front of him, at red lights, in the cycle stop area. He had small dick energy.

[Adam R @reason42]

@UK_Daniel_Card So I too am on my backup bike (it’s much better - gravel bike - love it)

[Adam R @reason42]

My temporary office for the day.

[Adam R @reason42]

@UK_Daniel_Card I might join you 😴🤣

[Adam R @reason42]

…at networking events even the waiters ignore me… 🥹

[Adam R @reason42]

@codewithava I got from 350 to 380 in 4 years. Growth comes slow for some 🤣🤞