If you’re not alternating Linda Rondstat and Ella Langley at full volume right now are you even alive?

LLMs in cyber is the natural progression. Aleph1 / Mudge published buffer overflows. Everyone freaked. Metasploit made writing exploits faster / more modular. Everyone freaked. Bindiff made 1 days nearly instantanous, everyone freaked. LLMs make it so you can think less about the exploit code, helps you narrow down exploits paths. Everyone freaks. Offense always accelerates in big jumps, then goes seemingly dormant (its not, just secretly monetizing) Blocking /censoring / regulating powerfull LLMs only hamstrings the defenders, not the attackers is my fundamental belief, mostly because that rule holds true in every other aspect of life. Guardrails and censorship and explot controls. Its almost as if people don't think blackhats are inside OpenAI and Anthropic networks right now as we speak. I always bet on the blackhats.

Go rewatch You Don't Mess with the Zohan.

The US is run by boomers for the most part and has been for decades. Boomers are fragile, easily scared, and non technical. They are also fundamentally non tolerant of diverse thinking. They are also rather power / control focused. All of this is regardless of political party. If you say a bunch of crazy shit like AI is as powerful as nukes, then they are going to try to control it and you will get to see what boomer central planning feels like. (Hint, it feels like curse words being censored on TV for appearances sake even though everyone curses all the time.) I’ve been very bullish on local models and anti guardrails for a very long time because of this exact situation which I saw coming. The US and Europe desperately need frontier lab style local models or else Chinese models will be the norm, or boomers will try to ban those too and we will be underground exchanging models like it’s PGP in the 90s during the Clinton anti cryptography era. None of my statement is politics which I care 0 for. It’s psychology, market dynamics, and a view on the current AI situation.

If prompting is criminal, only criminals will prompt.

I'm a free llm absolutist.

Food trucks please please get a website that tracks your current location. I’ll video code it for you if needs be. Schrodingers food truck isn’t working for me anymore.

Who in my network regularly gets around LLM guardrails?

If you get no credit for vibe coding something then the engineering lead also gets no credit.

Normies use "ChatGPT" the same way Texans use "coke" aka "What kind of coke do you want? Sprite, orange, dr pepper?"

What kind of breakfast did I just have? 1.) Cold, leftover bacon 2.) 2 pieces sourdough toast; 1 butter, 1 apricot habanero jam 3.) A buffalo meat stick 4.) Some Mrs Vickies chips with Anita's Loco Queso 5.) Some Kefir 6. Some popped water lily seeds

"A wave is the shadow cast by a rotating vector." Few high school math statements have affected me so deeply.

Quantum Cat says excuse me to the placement of angry octopus.

Mind Blowing

Dear businesses. I will never be using the mobile app. It’s not “my” mobile app it’s yours and I don’t want it.

I updated my debian box. Updating it broke my kerberos / NFS / automount setup. Things have gotten so complex there is no way to be sure updates won't break things. At the same time, exploits for vulns come out instantly so you have to update immediately. This is a very bad situation.

Ive spent a considerable amount of time building a binary analysis system for ELF, PE, MachO, etc. I intended that my use interface be a web console. Today it dawned on me I could just point my local LLM agents at the code base and tell it to use it, and improve as it goes, filling in gaps or fixing bugs. I took myself out of the loop.