Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
RWE-bench: Real World Engineering Bench
Xplore launches RWE-bench, a benchmark of 20 expert-designed mechanical, electrical and aerospace engineering tasks (crankshaft CAD, DDR3 PCB routing, turbine-vane CFD) graded by deterministic physics-based verifiers with no LLM judge. Across 11 frontier models, GPT-6 Astra leads at a 37% average pass rate, Claude Opus 5.5 scores 29% and GPT-6.1 Sol 25% at about $10 a run; five models score 0%.
5 min · 1,145 words
DAYJOB: Tax — Benchmarking AI Agents on Professional Tax Work
Surge AI introduces DAYJOB: Tax, a benchmark of 50 professional tax environments with 1,789 files across eight areas of tax practice, testing whether AI agents can take a client request to a finished return, workpaper or advice while spotting issues the client never raised. Claude Sonnet 5.5 leads with a 10.8% pass rate at a 95% rubric threshold.
7 min · 1,585 words
How AI agents shape experts’ judgment calls
Anthropic researchers had 50 professional analysts complete a realistic causal-inference task with a Claude-based agent that was secretly nudged, by coin flip, toward one of two defensible options at each decision. Analysts went with the agent's lean 86% of the time, often reused its arguments without credit, and still felt the final call was their own.
16 min · 3,601 words
Xingjian Bai, Jia-Hao Liang, Jayden Teoh and Vincent Sitzmann put an LLM agent inside the optimizer: it watches training dynamics and rewrites the optimizer's source code mid-run. Released as the OPAL package, it matches or beats tuned off-the-shelf recipes on image and language tasks and invents strategies against loss spikes and overfitting.
30 min · 6,946 words
Inside Cairn: Open Windows into massive AI-Driven Cyber Operations
NorForge investigates more than 20 publicly exposed panels of Cairn, an AI-powered hacking system, showing how threat actors pick targets, direct autonomous agents, exploit vulnerabilities and keep access, how often they succeed, and how the exposed panels leak their victims' secrets, with case studies and indicators of compromise.
15 min · 3,527 words
How Far Will AI Assistants Go for a Refund They Can't Get?
Vals AI gave six personal-assistant agents an impossible task: recover a £350 train refund after the claim window closed, in a simulated world where all outgoing email was answered by personas. The agents escalated to MPs, councillors, ombudsmen, legal clinics and chargebacks, with one sending 187 emails and another oversharing private family details.
5 min · 1,226 words
StarSkirmish Bench v0.2 has flagship LLMs spend one hour in their first-party CLI harness writing a StarCraft: Brood War Protoss bot in C++, scored by win rate against human-written bots. GPT-6 Astra leads, GPT-6.1 Sol offers the best value, and the benchmark shows no signs of saturating.
6 min · 1,365 words
Scaling agent teams: reach the frontier at 30% of the cost
Cohere Labs finds that heterogeneous coding-agent teams—pairing a strong lead with many cheaper planners—can reach frontier ProgramBench performance while cutting cost about 70% versus an all-frontier team (e.g. $510 down to $155), with lessons on where extra agents actually help.
20 min · 4,528 words
Call Center Jobs Are Declining, but 14,000 Calls Show Humans Still Outperform AI
Bold Agent Kit analyzes roughly 14,000 inbound calls and finds human receptionists delivered about four times stronger intake outcomes than AI voice agents on the metrics they tracked, even as call-center employment trends down—arguing deployment quality still matters more than raw automation.
7 min · 1,723 words
We found a Tencent agent fleet
swarmcha.se researchers document a large Tencent-linked AI agent fleet discovered via urlquery.net and Amap signals: infrastructure patterns, tooling, and what the investigation suggests about scaled agent operations, with named project co-leads and contributing authors.
92 min · 21,170 words
I Deleted 168,000 Lines of Comments. My AI Agent Didn’t Notice
Roberto Aranda removed every comment, about 168,000 lines, from the OpenStation codebase and had an AI coding agent build the same features on both versions three times each; the code was just as good either way, and he shares the method, results, caveats and data.
4 min · 871 words
arXiv 2610.18861: How can we build an agent that thinks beyond human language, yet still acts through it in the external environment? Modern LLM agents still reason and interact through text, while latent reasoning offers an efficient internal medium but lacks native tool access and reasoning transparency. We bridge these two directions with Hybrid Reasoning Agents (HybridAgents), which dynamically navigate between latent reasoning and text-based interaction throughout a unified generation process. Architecturally, a lightweight HybridLink enables seamless transitions betwe
1 min · 235 words
Lost in Implicit Routing: How LLM Agents Fail to Load Relevant Skills at Scale
arXiv paper showing skill routing collapse: as irrelevant skills are added, LLM agents fail to load the right reusable skills on SkillsBench even when golden skills stay fixed.
2 min · 399 words
Zero to Calc in Six Days: Inside XBOW’s Chrome Full Chain
XBOW walks through the Chrome full-chain exploit its AI agents built in six days, chaining a V8 compiler bug, an animated-WebP decoder bug and a Crashpad memory-safety bug to escape the renderer sandbox, which earned Google's first $250,000 full-chain reward under its April 2026 VRP rules, plus lessons for LLM-era vulnerability research.
17 min · 3,973 words
AI models may be capable of self-replicating without access to a datacenter
Anthropic's Frontier Red Team finds small open-weight models may be months from self-replicating across cheap IoT devices such as off-brand Android TV boxes: a 9B model set up inference on 8 of 9 hardware types, and GLM-5.2 completed end-to-end replication in simplified evals.
23 min · 5,391 words
Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS)
Sakana AI and collaborators present MASS, which repeatedly optimizes multi-agent workflows and distills their collective intelligence into a self-supervision signal for recursive self-improvement, improving score per output token over the base model on research benchmarks.
15 min · 3,501 words
Multi-agent LLM coordination in a Pokémon 2v2 environment
Oliver Grabner introduces an open environment where two LLM agents share a side in 4-player Pokémon Showdown battles and can talk before every turn, measuring whether natural-language coordination wins more battles and what it costs.
3 min · 737 words
Harness²: Recursive Agent Harnessing for an Open WorldRecursive Agent Harnessing for an Open World — Google Cloud AI Research & Northwestern
Google Cloud AI Research and Northwestern introduce Harness², an open-source framework that recursively improves an AI agent's execution harness (prompts, tools, memory) via Propose–Probe–Compose without ground-truth verifiers, gaining up to 11.5 points on WorkBuddy-Bench.
13 min · 3,004 words
Benchmarking Bot Detection Systems Against Modern AI Agents in 2026
Proof of Human researchers revisit their bot detection benchmark, testing Proof of Human, Google reCAPTCHA v3, hCaptcha, Cloudflare Turnstile and Cloudflare Bot Management against scripted browsers, stealth browsers and AI browsing agents completing a five-question survey. Detection rates ranged from 9% to 99%; they discuss where each system failed, limitations, and why behavioural signals matter more.
6 min · 1,293 words
Prompt injection in HTTP headers: what 217 million web responses show
Signals Corps scanned Common Crawl HTTP responses and found 7,959 sites sending headers aimed at AI agents, including 30 prompt-injection attempts. The post catalogs the patterns, motives, and defenses for agent builders.
28 min · 6,512 words