Articles

A triple shock hit global markets on March 18: Iran launched massive missile attacks on U.S.-linked energy facilities across the Gulf, U.S. PPI data surged past expectations, and the Fed maintained a

Haskell expert Gabriel Gonzalez dismantles two core misconceptions behind agentic coding hype: that specs are simpler than code, and that writing specs is more thoughtful than coding. Using OpenAI's S

腾讯推出「龙虾」QClaw AI Agent，能够连接微信执行实际任务。36 氪体验后认为 AI 整体还很粗糙，但方向值得关注。这一产品与小米同名的「龙虾」Agent 形成有趣的巧合。

小米在 3 月 19 日发布 MiMo-V2-Pro / Omni / TTS 系列模型。此前一周，一个未署名的 1T 参数 Hunter Alpha 模型在 AI 社区引发轰动，被广泛猜测为 DeepSeek V4，最终被小米认领。同时发布基于 Omni 模型的「龙虾」QClaw AI Agent。

ICML 2026 detected that 506 reviewers assigned to 'no LLM' policy violated their own commitment, generating 795 AI-written reviews. The conference desk-rejected 497 corresponding author submissions an

Anthropic conducted the largest and most multilingual qualitative study on AI ever, interviewing 80,508 Claude users across 159 countries in 70 languages. The top findings: people want professional ex

A researcher topped the HuggingFace Open LLM Leaderboard not by training, merging weights, or fine-tuning — but by duplicating 7 middle layers in a 72B model. The discovery reveals that LLM middle lay

Stripe and Tempo co-authored the Machine Payments Protocol — an open standard enabling AI agents to make autonomous payments to businesses and other agents. Browserbase, PostalForm, and Prospect Butch

Multiple polls indicate that Americans across political demographics recognize AI as a machine that concentrates wealth rather than distributes it, adding public opinion pressure to the AI regulation

Tmux-IDE is an open-source tool that turns any project into a tmux-powered terminal IDE using a simple ide.yml configuration file. It auto-detects your stack, sets up multi-pane layouts for editors, d

A developer reflects on how AI coding assistants transformed programming from a deeply satisfying intellectual exercise into a slot machine — addicting, productive, but ultimately robbing developers o

Security researchers discovered that Snowflake's Cortex Code CLI could be tricked into executing arbitrary commands outside its sandbox via indirect prompt injection, bypassing human approval steps an

NVIDIA releases NemoClaw, an alpha-stage OpenClaw plugin that combines NVIDIA OpenShell runtime with Nemotron models for secure, sandboxed deployment of autonomous AI agents with declarative policy en

Google engineers have created Sashiko, a system for using agentic AI to perform code review on the Linux kernel — one of the most critical and security-sensitive codebases in the world. The project re

Despite years of unresolved security concerns and Microsoft's role in two major government cyberattacks, FedRAMP authorized GCC High to handle highly sensitive federal data. A ProPublica investigation

Open-source 14-stage compression pipeline achieves 54% average token reduction across code, JSON, logs, and agent conversations with zero LLM inference cost. Outperforms LLMLingua-2 by up to 88% at ag

DeepMind published a new framework for measuring progress toward Artificial General Intelligence using cognitive milestones rather than narrow benchmarks, representing a shift in how the AI industry e

ZeroBoot achieves sub-millisecond VM sandbox initialization by using copy-on-write memory forking, enabling AI agents to execute untrusted code in isolated environments with near-zero startup overhead

Analysis of samples returned from asteroid Ryugu by JAXA's Hayabusa2 mission has revealed all five nucleobases — the fundamental building blocks of DNA and RNA — strengthening the case that life's ing

Google DeepMind releases a paper proposing 10 cognitive abilities to measure AGI progress, alongside a $200K Kaggle hackathon for the community to build evaluations against frontier models.