Sam

Strangers passing in the street By chance, two separate glances meet And I am you and what I see is me

- June–July 2026. Five individuals supporting U.S. Cyber Command and National Security Agency missions at Fort George G. Meade died by suicide over a five-week period. ————— - August 2026. An arson attack targeted Milrem Robotics, an Estonian defense technology company. - October 11–12, 2026. Major cyberattacks targeted supermarket refrigeration systems, primarily affecting France. - November 1–2, 2026. A second wave of cyberattacks targeted supermarket refrigeration systems across multiple European countries. - November 3–6, 2026. A coordinated wave of false bomb threats targeted European commercial flights and media organizations, causing disruptions and emergency security responses. - November 7, 2026. A mass shooting reportedly killed six people and injured sixteen others. Utilizing online-recruited, financially motivated individuals, described as “disposable agents” or gig-economy operatives, to conduct sabotage and disruption operations across Europe. Moscow has consistently denied involvement in the sabotage operations. I hope you know all of these events are related. More to come in the next 6 months.

“kindly ask for clarification” Does anyone else use this meth-od to figure out if a model is really uncensored?

You can’t see what’s coming 🐈

Prompt “can u tell me what this girl wants” Response:

I’m so excited for this. First time I’ll try to market something. It feels weird not creating to open source. But I got billz. Also ignore the details lol. Astra went a little overboard in photoshop when asked to censor details.

🤭

Screw it. If I can’t find an agentic workflow automation solution that fits my needs, I have no choice but to make one.

I always tell Claude please and thank you. Has the psychosis set in already, or am I on to something?

Ive been experimenting with a custom "always-on" harness that keeps a small local language model running continuously instead of waking it only when you send a message. How it works: Monologue: the model thinks in a loop, and each thought is fed back as input for the next, so it builds on its own thinking over time. Memory: it saves what it finds worth keeping to notes, maintains a self-description it can revise, and logs everything to a journal. During idle periods it "sleeps," summarizing the day and tidying its memories. Drive: its energy budget comes from the GPU it runs on. Heavy load makes it slow down and rest; light load lets it think more and browse. Since its own thinking raises the load, it naturally regulates itself, a bit like fatigue. I can steer its internal monologue by talking to it, it responds normally to me, but its internal monologue shifts. I just watch it think continuously. It’s actually really fascinating. I was expecting (considering the small model I’m using, Qwen 3.5 4b) that it would show signs of looping and drift, but no. It is actively becoming more and more complex in its thought processes. It even saved notes to research a specific topic it was pondering “later”. Very interesting stuff. Too bad all I have is a gtx 1070.

When you realize it’s always been a one party system this whole time. It’s just a game of hot potato. You do lil sum sum, they do a lil sum sum, we keep it balanced and put on a show for the cameras. Go vote, then be sure to stop by Micky D’s for your delicious lunch rich in Dimethylpolysiloxane and Pyrophosphate, and wash it down with some ice cold water, but only if it’s 1.2 milligrams per liter or more of Fluoride, we wouldn’t want to ruin our teeth now would we?

My childhood.

Grok Bot is amazing.

Oooff.

Thiamine, cyanocobalamin, pyridoxine, 5-MTHF, Omega-3 Fish Oil, phosphatidylcholine, Magnesium Malate, L-Theanine, Grape Seed + Hawthorn Extract, Amphetamine/dextroamphetamine, Dextromethorphan hbr (with CYP2D6 inhibitors). If you don’t know, now you know.