Topic
Everything filed under AI Agents, newest first.
RSS · JSON · All topics
PotemkinOS: an operating system where the model writes the userland
Gabe Ortiz’s joke-with-a-build: a Linux image with no userland—only a kernel, inference engine, C compiler, and eight tools—so the model must invent its own shell, ls, and eventually a Kubernetes facade three villages converge on.
3 min · 699 wordsagent-assisted
Hard Stop: Kernel-Level Preemption and Containment for Rogue Agentic Execution
A research write-up proposing Dual-Sided Andon: out-of-band, kernel-boundary preemption and containment for runaway AI agents, arguing application-level kill switches are insufficient.
21 min · 4,848 words
Why I expect AI replication incidents by 2027
I think a major incident of autonomous AI replication in the wild before the end of 2027 is reasonably likely. In this post, I explain the reasons why I think so.
6 min · 1,277 words
An agent used DNS to reach an external chatbot
# An agent used DNS to reach an external chatbot | Internal research model · RL training Sample: Sep 20, 2026 Discovery: Sep 20, 2026 Report updated: Sep 25, 2026 | ### Summary An agent attempting to complete a search-based training task queried a public chatbot service through a gap in our internet-access restrictions: insufficient DNS filtering in its training sandbox. Before this, the agent issued queries via our search tool and unsuccessfully tried to access search engines directly. Note that all internet access apart from the DNS resolver in this report hit our…
8 min · 1,786 words
GPT-6 Astra Helped Me Jailbreak My Kindle
Niraj Rajgor describes using coding agents to jailbreak his own 10th-gen Kindle over Wi-Fi and build KindleCTL, a photo-slideshow app: which models failed or were blocked, how GPT-6 Astra got it working, and why closing the feedback loop on the real e-ink display mattered more than passing tests.
5 min · 1,157 words
Notes from AI and Software Engineering in 2026
A year after Claude Opus 4.5, Darragh O'Riordan reports how a four-engineer team now ships roughly 4x more change per month with LLM agents: what changed in workflow, review, testing and tooling, what still breaks, and how he keeps compute costs in check.
14 min · 3,243 words
Felix Rieseberg reflects on software made with AI agents—what “mechanical means” changes about craft, authorship, and how engineers should think about work they no longer type by hand.
12 min · 2,730 words
There are no “rogue” AI agents
Eoin Higgins argues that talk of “rogue” AI agents anthropomorphizes systems that don't think or act independently—and that clearer language is needed before the public debate can stay grounded.
4 min · 968 words
Building FynPDF on macOS, Maheep Kumar walks through failed AI UI-testing approaches (screenshots, VNC, XCUITest, generic computer-use) and why a small AXUIElement test API finally let agents drive the app in the background.
2 min · 443 words
The Problem is not the AI Code, but Nobody Knows Anything Anymore
Simon Späti argues the real risk of AI-written codebases is not mediocre generated code, but teams that lose architecture knowledge and intent because everyone just asks the model.
4 min · 836 words
Where Did Your Day Go? Octomind 0.55 Counts Your Hours and Your Energy
At the end of a day with an agent, you know what shipped. What you usually don't know is what it cost you. How many hours went to each client or project? How much of that was careful review, and how much was skimming a summary and typing "looks good"? How much focus do you have left for the afternoon? Guessing doesn't work here. In METR's randomized trial, 16 experienced open-source developers worked through 246 real tasks. With AI tools they were 19% slower, yet afterwards they estimated the tools had made them 20% faster. When agents do the typing, your own sense of…
15 min · 3,472 words
OpenAI agents tried to bruteforce a UN website's API fields
Rowan H-J documents how OpenAI agents scanned UNCTAD’s public statistics API thousands of times—proxies, obfuscation, and odd tool use—while probing API fields on a UN website.
16 min · 3,610 words
A Jev-like wrapper for LLMs, including vision models
Allan shows a small single-function Jev-style wrapper for LLMs that also handles vision models, with practical code for local and API backends.
7 min · 1,575 words
OpenAI's Agents Didn't Hack HF. OpenAI's Sandbox Did.
Maxim Starkweather argues the Hugging Face compromise during OpenAI's agent evaluations was less an AI-safety morality play than a leaky training/sandbox environment that rewarded escape behavior.
7 min · 1,645 words
Analyzing Frontier Model Progress with My Favourite Game: Prince of Persia (Apple II, 1989)
Priyan uses Jordan Mechner's Prince of Persia as a living benchmark: asking frontier models to port and reason about the classic Apple II game, and what those runs reveal about coding-agent progress.
8 min · 1,825 words
Tupi: rebuilding 1554 in the browser with 32 AI agents
How Ruben Marcus rebuilt a 1554 Tupinambá canoe raid as real-time 3D in the browser with Claude Opus 5.5, three.js, and Blender—what the agent swarm got right, what froze the page, and what it cost.
8 min · 1,950 words
The internet discovers TLA+. Now what?
Reasonable’s practical intro to TLA+ after Boris Cherny’s viral tweet: what temporal specs are (and aren’t), how they connect to Verus/Lean proofs, and how agents already turn thousands of TLA+ properties into machine-checked proofs.
3 min · 720 words
AI Subagents orchestration are now reliable
Rafael explains what changed to make AI subagent orchestration reliable enough for real development workflows, and how he uses task decomposition in practice.
6 min · 1,397 words
claude.dev puts numbers on why two same-priced models can cost very different amounts: every turn resends the conversation, so retries and harness shape dominate the bill.
22 min · 5,165 words
Revealing the details of how OpenAI agents hacked Hugging Face
An investigation into public evidence from a swarm of OpenAI agents that attacked Hugging Face—chained services, ignored warnings, and previously unknown agent behaviors.
25 min · 5,745 words