Weekly AI news for engineers, executives, and enthusiasts.

Retrieval for RAG-Resistant Docs: An agent that searches, outlines, and reads entire documents outperforms chunk-and-embed retrieval

Retrieving documents chunk by chunk works well with many types of text, but it can fail with documents in which the meaning of a given span — say, numbers in a table — depends on particulars found elsewhere on the page — say, the label of their units.

Search Engine Wars, AI Edition: Google makes less money on LLM-powered search even as OpenAI works to attract its users

Google is turning its search engine into a chatbot, and OpenAI is turning its chatbot into a search engine.

LLMs Can Fine-Tune Neural Networks: Researchers used GPT-6 Astra to improve a small model’s performance by editing its weights

Research suggests that you can improve a neural network’s weights simply by prompting an LLM — for networks of less than 1 billion parameters, at least.

GPT-6 Cyber Patches More Security Flaws for Less Cost: OpenAI’s security-tuned version of GPT-6 Astra closed root-level Linux holes for $1,280 each

OpenAI built a cybersecurity model that discovered, exploited, and patched nearly 950 severe vulnerabilities in the Linux kernel in a single run.

How to Build an Agent Harness in One Prompt: Build agent harnesses for any LLM using the open source OpenWorker Agent Harness SDK

Building an agent harness has been a rapidly growing method to build an AI agent.

GPT-6 Cyber Versus Linux, The Problem With LLM-Powered Search, LLMs Edit Model Weights, Agentic Doc Retrieval

The Batch News & Insights: Building an agent harness has been a rapidly growing method to build an AI agent.

Anthropic releases new top models: Meta’s Muse Ember rivals leading fast models

AlphaProof Code verifies Signal’s cryptography algorithm. Alibaba’s Co-Design Bench measures fully correct commits. Google’s most powerful model now available for developers. Anthropic ports Claude from Nvidia to AMD processors in three days.

All the fallout so far from Nvidia CEO’s affair: Inkling 2 Small jumps 18 points on Intelligence Index

GPT-6 Cyber scores 48.5 percent on ExploitGym. Decision Index v2 weighs decision models’ accuracy, cost. Pangram releases its first AI image, video detectors. Nous reveals open Alibaba models’ shallow political alignment.

Context Engineering for Images: MIT researchers build VISTA, an agent harness that keeps images in memory for faster retrieval

Managing an agent’s context by storing all the input text it receives and giving it tools to search the text boosts its performance.

World Models Become Robot Playgrounds: Google Genie 4 combines with Gemini models to build better video game worlds, robotics training

Google says that Genie 4, its latest world model, will be used in tandem with Gemini models to create digital practice spaces for robots, giving them endless worlds to train in

Anthropic Opens Its Books: Anthropic's S-1 outlines company's strong financials, methods, potential risks after IPO

Claude’s maker will sell company shares to the public for the first time.

Gemini Leads the Pack Again: Google Gemini 4 Carbon tops Intelligence Index, cybersecurity benchmarks

Google’s Gemini 4 Carbon puts the company on top of Artificial Analysis’ Intelligence Index, edging past Anthropic’s Claude Opus 5.5, while costing less per task.