AI governance expands; China bans Claude

Anthropic / Claude ecosystem

Claude Code Weekly Limits Extended 50% Through August 19

Anthropic has extended the 50% higher weekly usage limits for Claude Code through August 19 for all Pro, Max, Team, and seat-based Enterprise users. This follows an earlier temporary limit increase that was set to expire on July 19.

Frontier model providers

OpenAI Lets Parents Mandate Study Mode for Teen ChatGPT Sessions

OpenAI has launched a parent-controlled Study Mode for teen users of ChatGPT that teaches step-by-step problem solving, along with expanded safety notifications. The feature allows parents to mandate educational use and limit general chat.

AI developer tooling & infrastructure

LM Studio Launches Bionic, an Open-Model Coding and Document Agent

LM Studio has released Bionic, an AI agent designed for coding and document tasks that runs on open-weight models. Users can control inference location and costs, making it a flexible alternative to cloud-based agents.

Cloud & platform providers

No significant new developments.

AI policy, regulation & governance

Chinese State-Owned Firms Ordered to Replace Claude Code with Domestic AI Tools

China has ordered state-owned enterprises to replace Anthropic's Claude Code with domestic AI alternatives, according to reports. This move is part of an expanding campaign to tighten control over foreign AI software usage in key sectors.

Open-Weight Models Narrow Cyber Capability Gap to Four Months Behind Frontier

A report from the UK's AI Security Institute (AISI) finds that open-weight models like DeepSeek V4-Pro and GLM-5.2 are now only four to seven months behind frontier systems in offensive cyber capabilities. The research highlights a rapidly shrinking gap.

EU Adopts AI Governance Principles for Health Technology Assessments

The EU Health Technology Assessment coordination group has adopted governance principles for using AI in health technology dossiers, mandating transparency and human oversight. This guidance applies to joint clinical assessments.

White House Dictating Access to Frontier AI Models

The Biden administration is reportedly negotiating with AI developers to control access to frontier models, shifting leverage from tech giants. This move aims to manage risks and ensure national security.

EU Compels Google to Share Android and Search Data with Rival AI Assistants

EU regulators have mandated that Google share Android and search data with competing AI assistants, aiming to boost competition. This decision could reshape the AI assistant market in Europe.

Industry & market moves

SpaceX Secures Billion-Dollar Pentagon AI Compute Deal

SpaceX has secured a multibillion-dollar deal to provide AI compute capacity to the Pentagon, marking a significant expansion from space launch services into defense AI infrastructure. The deal could reshape the defense-AI landscape.

AI product & feature launches

Microsoft Project Perception Aims to Rival Mythos for AI Bug Hunting

Microsoft has announced Project Perception, an AI bug hunter that uses multi-model vulnerability detection at lower cost than Anthropic's Mythos. The tool is expected to be available on July 31, 2026, with wider access.

Nonprofit Current AI Building Open AI Infrastructure, Free for All

Current AI, a newly formed nonprofit, is racing to build open, public AI infrastructure as an alternative to corporate-controlled AI. It has secured government partnerships and is releasing open-source AI devices.

Aether-7B-5Attn: Fully Open-Source Sovereign Foundation Model Released

A new open-source model, Aether-7B-5Attn, has been released with a byte-for-byte reproducible training recipe and logs. It serves as a sovereign foundation model for organizations seeking full control.

Alibaba Launches Qwen 3.8 Open-Weight Model with 2.4 Trillion Parameters

Alibaba has released Qwen 3.8, a 2.4 trillion-parameter open-weight model that claims near-frontier performance, second only to Anthropic's Fable 5. The model is available for commercial use.

Nubia Announces AI Agent Smartphone NaviX Ultra with On-Device AI

Nubia has unveiled the NaviX Ultra at WAIC 2026, claiming it as the world's first AI agent smartphone. The device features on-device AI processing, potentially enabling autonomous task execution.

ACE Robotics Unveils Kairos 3.1 Embodied AI Stack at WAIC

ACE Robotics introduced Kairos 3.1, an action-oriented world model, and the Ambient Capture Engine 2.0 at WAIC 2026, expanding its embodied AI stack from data collection to deployment.

Research with immediate practical relevance

VideoChat3 Open-Source Model Beats GPT-5 on Video Grounding

VideoChat3, a new open-source model, has outperformed GPT-5 on video grounding tasks. The full training stack, including weights and datasets, has been released, enabling reproducibility.