本周 AI 回顾 — 2026年7月27日 – 8月2日

Google DeepMind announced that its latest AI model, Gemini Robotics 2, can now control the entire body of humanoid robots. This new version supports "whole-body motions" from feet to fingertips, a significant advancement from the previous model which only focused on controlling the upper body. This development marks a step forward in robotic control capabilities.

Google announced that its internal AI tools helped patch more security flaws in the Chrome browser in June than in the past two years combined. A chart published by Google, as part of a white paper on using AI to find and fix flaws faster, illustrates this exponential increase. Chrome’s 126 was released in June 2024, with Chrome 149 and 150 released last month, each version being a "milestone."

The Gemini Robotics team developed "Gemini Robotics 2," a system designed to bring whole-body intelligence to robots. This initiative involved a large team of researchers and engineers, including Abhijit Ogale, Abhishek Jindal, Adil Dostmohamed, and many others from DeepMind. The project aims to advance robotic capabilities by integrating sophisticated intelligence across the robot's entire physical structure.

OpenAI has released the GPT-5.6 model series, including Sol, Terra, and Luna, within the Kiro software development agent. This update aims to enhance AI-native coding by enabling developers to generate higher-quality code with fewer iterations and increased value per token. OpenAI and AWS collaborated to optimize the Kiro environment and OpenAI models, with tests showing GPT-5.6 Terra reducing successful task costs on Terminal-Bench 2.1 by approximately 82%. Kiro's specification-driven approach allows models to find working solutions faster, leading to more completed work and greater value for developers.

OpenAI has open-sourced @openai/codex-security, a CLI and TypeScript SDK designed to identify, validate, and remediate security vulnerabilities in code. This tool allows users to scan repositories, review changes, track findings, and integrate security checks into their CI pipelines. It requires Node.js 22+, Python 3.10+, and access to Codex Security, with detailed documentation available for setup and usage.

Kimi-K3 on HuggingFace presents various benchmark results, including Coding and Agentic capabilities. For Coding, benchmarks like DeepSWE, ProgramBench, Terminal-Bench 2.1, FrontierSWE, SWE-Marathon, PostTrainBench, MLS-Bench-Lite, SciCode, and Kimi Code Bench 2.0 are listed with their respective scores. Agentic capabilities are evaluated using BrowseComp. Additionally, PerceptionBench is mentioned as an in-house benchmark focusing on atomic visual perception capabilities.

A local, zero-cost merge queue for parallel Claude Code agents is introduced, designed to serialize agent operations to prevent push races, redundant heavy builds, and shared-resource test flakiness. The system supports configuration options such as `branchPrefix`, `worktreeSuffix`, `portBase`, `integrationBranch`, and `productionBranch`. It also allows defining `protectedBranches`, `regenerableFiles`, `symlinks`, `buildOutputDirs`, and a `checkCommand` for gating landings, with `checksRequired` to enable or disable these checks. The project is open-source under the MIT license.

ORCA-bench is a new benchmark designed to evaluate language model agents in a production-fidelity oncall setting for root cause analysis (RCA). It uses a live OpenTelemetry-instrumented microservice system with six days of metrics, logs, and traces, and 1,079 RCA tasks. Expert SREs curate ground-truth symptoms. The best agents achieved only 25.3% RCA Accuracy on Medium-difficulty tasks and 10.0% on Hard tasks, even with Claude Fable 5. This indicates a significant gap before these agents can be safely entrusted with production reliability.

Scientific computing is crucial for modern research, but its software often lags due to origins in academic teams with limited engineering. This leads to slow, fragile workflows that hinder discovery. Agentic AI, like GPT-5.5, can accelerate this by modernizing tools; for example, GPT-5.5 improved cyvcf2's build and packaging, making it easier to install, test, and release. As coding agents advance, researchers can focus more on discovery and less on maintaining analysis pipelines.

Anthropic has published details of a practical key-recovery attack targeting HAWK-256. This development, related to AI CODE CREATION, highlights potential vulnerabilities in cryptographic systems. The information was shared within the developer community, indicating its relevance for those involved in enterprise-grade security and premium support, emphasizing the ongoing need for robust cryptographic solutions.

Noisegate is a differential-privacy gateway designed to provide AI agents with query access to sensitive data, ensuring that no individual's record can leak. It achieves this through a trusted privacy engine that clamps data to declared ranges, adds calibrated Laplace noise, and decrements a budget. The system includes a validation/guardrail layer for policy checks and handles untrusted LLM compiler output by emitting a CONSTRAINED QUERY AST, not free SQL, via structured/schema-forced output.

OpenAI is committed to advancing responsible AI in Europe, aligning with the EU AI Act. They emphasize safety, security, transparency, and provenance, contributing to the EU’s General-Purpose AI [GPAI] Code of Practice and the Code of Practice on Transparency of AI-Generated Content. OpenAI provides resources like model documentation, system cards, and usage policies to help customers and developers prepare for the Act's implementation, aiming to maximize AI's benefits while managing risks.

The OlmoEarth Platform, detailed at allenai.org/olmoearth, enables geospatial inference at a planetary scale. Its independent partitions allow stages to run across thousands of compute instances simultaneously. This was demonstrated by generating a North American wildfire-risk map, utilizing 19,600 CPUs and 994 GPUs in parallel, achieving a 155x speedup by reducing 4,737 hours of serial compute to 30.5 hours. The platform also aims to lower barriers to geospatial model use through agentic tools and interfaces.

AI Mode in Search can help users host dinner parties by integrating with apps like Canva. Users can prompt Search to "Create a flyer in Canva for the party I’m hosting in two weeks. Make it modern minimalistic style," generating a custom design directly in search results. This allows for easy editing or downloading, enabling users to focus on other party preparations.

Google Search can help with dinner party planning, from creating menus to discovering dishes, allowing hosts to focus on guests. This tool aims to simplify the heavy lifting involved in organizing a dinner party, ensuring a more enjoyable experience for everyone involved. The official blog post from blog.google, published on July 28, 2026, highlights "5 ways to host the ultimate dinner party with Google Search."