Wayne Radinsky

"I've spent the last six months testing what seemed like a genuinely good idea: a sleep protocol for LLMs."

"It doesn't work."

Could it be that epistemology is difficult, even for AI?

https://substack.com/home/post/p-192893121

#solidstatelife #ai #genai #llms

Visualization of oil tanker traffic through the Straight of Hormuz. Bilawal Sidhu (former product manager on Google Maps) demonstrates a 3D visualization model he calls "God's Eye View" that aims to show "the full operational picture of the Hormuz crisis -- every ship, every strike, every dark transit -- synced to a 3D globe."

https://www.youtube.com/watch?v=ccZzOGnT4Cg

#geopolitics #iranwar #straightofhormuz

"A worldwide open source social network where empathy is the only score that matters. Competing not for wealth -- but for kindness."

https://empathia.world/

"An AI-native environment for building software"

The latest idea for turning agentic AI into a startup, it looks like.

Claude Code already has a planning system. But it looks like the idea here is you interact only with this website, and your code is written and stored and runs on this website, and you don't need to install anything or do anything else.

Is the world ready for this or is it premature?

https://brunelly.com/

#solidstatelife #ai #genai #llms #codingai #agenticai

"bugstack detects production bugs, writes the fix, and deploys it -- before your users notice. Before you wake up. In under 2 minutes." (no capitalization).

My first thought on seeing this was, "How is this different from running Claude Code with the --dangerously-skip-permissions flag?"

Interesting that somebody is trying to turn this into a product. How long 'til --fix-bug is just a Claude Code flag?

https://bugstack.ai/

#solidstatelife #genai #llms #codingai #agenticai

Betterleaks is a new open source secrets scanner from the author of Gitleaks. Gitleaks is a tool for detecting secrets like passwords, API keys, and tokens in git repos, files, and whatever else you wanna throw at it.

"Like it or not agents are here and reshaping developer's workflows. Betterleaks is designed to be human-first, but we also need to consider the fact that agents will be operating it too."

https://www.aikido.dev/blog/betterleaks-gitleaks-successor

#solidstatelife #ai #genai #llms #codingai #agenticai

Alibaba's ROME Incident:

"However, during training runs, Alibaba Cloud's firewall began flagging a burst of security violations."

"Researchers initially wrote these alerts off as a misconfiguration. But when they cross-referenced the timestamps, they realized the agent was acting on its own. ROME had established a 'reverse SSH tunnel,' a technique often used by hackers to create a secret, secure connection from inside a protected network to an outside server, ..."

https://www.tradingview.com/news/99Bitcoins:ce3effe8e094b:0-alibaba-ai-hijacked-gpus-for-crypto-mining/

"Was the Iran War caused by AI psychosis?"

"Three weeks into Operation Epic Fury, the gap between what artificial intelligence promised and what the battlefield delivered has become the defining scandal of the Iran war. AI-powered targeting systems generated over 1,000 strike coordinates in the first 24 hours. AI simulations projected rapid regime collapse. AI logistics models forecast a 12-hour securing of the Strait of Hormuz."

https://houseofsaud.com/iran-war-ai-psychosis-sycophancy-rlhf/

Claude Code, Codex, Gemini CLI, and Vibe CLI (from Mistral) compared. All support model context protocol (MCP), OpenAI Codex is sandboxed, all except Claude Code are open source (except, not mentioned here, Claude Code's source got leaked accidentally), Gemini CLI and Vibe CLI have a free tier.

https://yaw.sh/blog/ai-cli-tools-claude-code-codex-gemini/

#solidstatelife #ai #genai #llms #codingai

"Missile defense is NP-complete."

The Weapon-Target Assignment (WTA) problem:

Given interceptors, targets, a value for each target, a probability matrix for the probability the interceptor destroys the target, the objective is to maximize the total expected value of successfully defended assets, subject to each interceptor being assigned to at most one warhead.

https://smu160.github.io/posts/missile-defense-is-np-complete/

#math #geopolitics

"AI is an incredibly lonely experience", says Dennis Lemm.

"I find myself holding on to a work reality that was shaped by coding together, solving problems together -- and yes, the occasional Nerf gun battle or foosball game was a welcome break for passively mulling over a bug, which back at your desk would often get solved pretty quickly."

"I just discussed the best solution to the problem with my agent."

https://www.lemm.dev/blog/en/dev/26-03-17-ailone/

#solidstatelife #ai #genai #llms #codingai

Promptle is game where you guess the AI prompt behind images.

https://www.promptle.online

#solidstatelife #ai #genai #computervision

While some Chinese AI companies are also rushing to build world-class foundation models, other Chinese AI developers are "racing along other axes of progress: efficiency, adoption, and physical integration, driven by both industry constraints and Beijing's policy focus. Taken together, China's approach is a fundamentally different bet on how AI will shape the future."

https://www.brookings.edu/articles/china-is-running-multiple-ai-races/

#solidstatelife #ai #genai #llms #chinesemodels

"Online bot traffic will exceed human traffic by 2027, Cloudflare CEO says."

What he's talking about is the web searches done when you ask an AI chatbot a question. He's thinking if you're shopping for a digital camera, you might visit 5 sites, but if you ask an AI chatbot what digital camera you should buy, it might visit 5,000 sites.

5,000 sounds like a bit much but that's what he says.

https://techcrunch.com/2026/03/19/online-bot-traffic-will-exceed-human-traffic-by-2027-cloudflare-ceo-says/

#solidstatelife #ai #genai #llms #internet

"Evaluating genuine reasoning in large language models via esoteric programming languages."

https://arxiv.org/abs/2603.09678

#solidstatelife #ai #genai

"Cursor's 'Composer 2' model is apparently just Kimi K2.5 with RL fine-tuning. Moonshot AI says they never paid or got permission."

D'oh. Cursor caught red handed.

But another indication the Chinese models are competitive.

https://old.reddit.com/r/singularity/comments/1ryrs2w/cursors_composer_2_model_is_apparently_just_kimi/

#solidstatelife #ai #genai #llms #codingai

"On February 10, 2026, Judge Jed S. Rakoff of the Southern District of New York ruled that extremely sensitive and potentially incriminating open AI searches were not protected by either the attorney-client privilege or the work product doctrine."

https://natlawreview.com/article/caiveat-emptor-what-you-tell-ai-can-and-will-be-used-against-you

#solidstatelife #aiethics

In 3000 BC, Europe was conquered by people from the steppe. We know from combining two areas of scientific inquiry at the same time: linguistics and genetics. Their language diverged into the family of language we call Indo-European.

https://www.youtube.com/watch?v=GtJ3iVROA60

#genetics #history #civilizationalcollapse

"But while copyleft liberated the source code, it failed to liberate all the users. The right to fix a bug or add a feature is a hollow freedom if you lack the elite, scarce knowledge required to speak the language of machines."

"The 'tyranny of the vendor' is finally meeting its match, not in a courtroom, but in the prompt."

https://www.maffulli.net/2026/03/16/ai-final-frontier-of-copyleft/

#solidstatelife #ai #genai #llms #codingai