Mine the Gap: Open-Source Tools for Measuring the AI Offense-Defense Gap | Dreadnode Dreadnode ran red team and blue team agents in the same Active Directory environment creating a realistic, closed-loop evaluation. Run your own evals with DreadGOAD and Ares.

👀 Mine the Gap: Open-Source Tools for Measuring the AI Offense-Defense Gap | Dreadnode m.cje.io/4mxNV3P

Me, when the teacher kicks out my bff and gives the seat next to me to the smart kid

Bro, I log on to this boring-ass social network precisely to see your Xi Jinping thought memes

Very excited for this book! Just preordered

Not to mention that they called in Soviet tanks to literally crush labor protests and strikes! They mismanaged the economy and then increased working hours by 10% at same pay. The Soviet troops and GDR police murdered more than 50 people and jailed thousands, in order to force factory work to resume

Mistake is to think of states as independent entities. If America were to default on its debt the consequence would be eliminating the retirement portfolios of a very large number of its citizens. So they will stop it. In the democratic world the citizens are the holders of sovdebt and the sovereign

Dreadnode — AI Infrastructure for Security Agents Dreadnode provides the infrastructure for building, evaluating, and deploying security agents.

Want to have some fun building agents?

Go to dreadnode.io

I never claimed that my encounters have been pleasant. I was commenting on the systemic setup in which TSA are not exploited laborers but fairly paid government employees. People having decent jobs is preferable to me. You inferred something I did not say and completely ignored the actual argument.

I disagree. I like the TSA. They are trained, unionized, actual government employees, usually amply staffed, and efficient. Compared to Germany, where airport security is privatized corporate security guards squeezed for margins by private companies and thus on strike a lot - TSA is a dream

This reads very much like it was written with Claude. The entire "it‘s not this… it‘s that" - the constant negation is one of the most telltale and most annoying signs of bad AI writing.

Final chemotherapy infusion is done! I'll still be recovering for a while, but I'm so excited to be wrapping up this phase of my treatment plan! Two weeks off, then I start radiation therapy!

Philip ❤

"You used to beep me on my pager. Late nights when you want to detonate. I know when that hotline bling, that can only mean one thing (my thumbs detaching)"

*Spinning a chair around backwards, sitting down with a group of first year Master's students at the pizza social*

"You know, real cyberwar has never been tried."

The way he started smiling 😊

To learn more about this period, read "Die Tante Jolesch"

Tutor: How long was the 30 year war?

Archduke: 5 years

T: Brilliant, Sire. Since they only fought during the day and not at night, 30 years become 15. Then discounting public holidays, rest days and extended negotiations we arrive at 5 years.

Though I wonder how good the coffee in the US is?

When Viennese intellectuals were disenfranchised by an autocratic system in the 19th century they would gather in coffee houses and play the "Archduke game". One would play an idiotic royal and the other his tutor. Every deliberately dumb answer had to be explained as a stroke of genuis

Turns out AI is a strategic studies major at Johns Hopkins SAIS

Signed up! Sounds very interesting!

Wait. So the US let Assad walk all over the red line without consequence. This was bad. But enforcing a red line is an excuse? Just because they learned from it?

When Assad killed civs not reacting was very controversial within the Obama admin. This admin has none of their qualms (or their ideals)

Hat der Open Technology Fund der US-Regierung für andere Content-Einschränkungen ja schon lange gemacht (TOR, Signal, etc). Nach chinesischem/russischem/iranischem Recht auch strafbar. Aber leider wurde das eingestellt und stattdessen das hier aufgenommen.

I‘ve been to this exact spot and I didn’t even know it was THE SPOT WHERE BRUCE MET ADAM! I huddled here after coming out of the Rodin museum and discovering it was pouring rain just after APSA had wrapped up

From 0 to DA: Inside Worlds, a Simulation Engine for Agentic Pentesting YouTube video by dreadnode

Want to build an AI agents that hacks stuff? Don‘t have training data? No problem! Find out how:

www.youtube.com/watch?v=9C7p...

Worlds: A Simulation Engine for Agentic Pentesting An 8B model went from blindly loading Metasploit modules to achieving Domain Admin on GOAD, trained entirely on synthetic data from our world model system.

We fine-tuned an 8B model to pop a GOAD domain…using only synthetic training data. No real networks. No frontier model distillation. Just a world model that simulates AD environments and generates realistic pentesting trajectories.

See how we did it: dreadnode.io/blog/worlds-...

Seems plausible but I‘d like to see a Granger test…

We about to have some Llama Drama :)

Screenshot of Snowstorm hitting east coast

The @districtcon.bsky.social gods ask for their due. The conference may proceed but one natural crisis will occur

I have 10gigabit symmetrical here in Switzerland