Business
- OpenAI 最强 AI 模型：GPT 5.6 系列正式上线，纳德拉称微软 Copilot 同步接入 新浪网：OpenAI has confirmed that its newly released GPT 5. 6 will be the 'preferred model' for Microsoft Copilot 365, pushing back against breakup speculation. The statement comes amid reports that Microsoft is exploring AI model diversification. Source: news.google.com、techcrunch.com
- Paris AI Voice Startup Gradium Raises $100M Seed Round Backed by Nvidia：Paris based AI voice startup Gradium announced a $100 million seed extension round backed by Nvidia and other investors. The company positions itself as a direct competitor to ElevenLabs in the AI voice synthesis market. Source: techcrunch.com
- AI agent startup Lyzr lets its own agent run its $100M Series B — a meta proof of concept：Lyzr, a three year old enterprise AI agent startup, used its own AI agent to lead its $100 million Series B at a roughly $500 million valuation. The company received $400 million in investor interest from Silicon Valley, the Middle East, and financial institutions, far exceeding its target. Source: techcrunch.com
- OpenAI's No. 2 Executive Fidji Simo Steps Down as Company Eyes IPO：OpenAI's second in command Fidji Simo is stepping down from her full time role after an extended medical leave, creating a leadership vacuum at a critical juncture as the company prepares for a potential IPO and races to catch Anthropic in the enterprise market. Her departure adds uncertainty to OpenAI's executive team stability during a pivotal growth phase. Source: techcrunch.com
- Zhipu AI Plans to Raise 31.41 Billion HKD Through Placement for Three Strategic Directions：Zhipu AI (Zhipu) plans to raise 31. 41 billion HKD through a share placement, with funds directed toward three major strategic areas. The massive fundraising move signals a major capital market action by one of China's leading AI companies. Source: news.google.com
- YTO Express launches six full stack AI achievements, marking large scale AI deployment in China's courier industry：YTO Express, one of China's leading courier companies, officially unveiled six full stack AI implementation achievements covering core logistics operations. The move signals that AI adoption in China's express delivery industry is moving from pilot programs toward large scale deployment. Source: news.google.com
- Deutsche Telekom partners with OpenAI to become an AI native telecom operator：Deutsche Telekom is deepening its partnership with OpenAI to embed AI across customer service, employee workflows, network operations, and voice interactions. Europe's largest telecom operator is pursuing a full AI native transformation. Source: openai.com
- MiniMax Announces 16 Billion HKD Funding Plan, CEO Yan Junjie Pledges Personal Salary and Shares for AGI：Chinese AI startup MiniMax has quickly announced a 16 billion HKD funding round following a market adjustment, with founder Yan Junjie pledging his personal salary and shares to pursue AGI. The massive funding plan signals sustained strong capital interest in China's AI sector. Source: news.google.com
Models
- Meta's Muse Image AI Generator Uses Public Instagram Photos by Default, Sparking Privacy Backlash：Meta's new Muse Image AI generation feature allows users to create AI images using photos from public Instagram accounts, raising serious privacy concerns. Users must manually toggle off the "Allow people to create with and reuse your content" setting to opt out. Source: techcrunch.com
- SpaceX Achieves Multiple AI Breakthroughs With AI Coding Tool Cursor：SpaceX has made significant AI technology advancements with the assistance of the Cursor AI coding tool, according to a report from Sina Finance. The development highlights the growing role of AI coding assistants in cutting edge aerospace engineering. Source: news.google.com
- New York Times Says OpenAI Hid Evidence in ChatGPT Copyright Trial：The New York Times and other news publishers have accused OpenAI of hiding tools and datasets that could identify copyrighted journalism in ChatGPT outputs. The plaintiffs filed a new motion for sanctions, escalating the high profile copyright lawsuit. Source: techcrunch.com
- Anthropic Develops "Jacobian Lens" to Reveal Claude's Hidden Reasoning Space：Anthropic researchers have built a technique called the Jacobian lens that offers the clearest view yet of what happens inside large language models as they process questions and tasks. The findings range from the mundane to the unsettling, showing how Claude puzzles over concepts before arriving at answers. Source: technologyreview.com
- Gege AI launches billion parameter Mandarin music model with ByteDance revenue sharing deal：Gege AI released a from scratch trained billion parameter end to end Mandarin music generation model that produces full 3 minute stereo songs with separate vocal and accompaniment tracks. The company also signed a copyright revenue sharing agreement with ByteDance, enabling AI generated music to be monetized across Douyin, CapCut, and other platforms. Source: qbitai.com
Infrastructure
- Meta's Next Gen AI Chips to Begin Production in September, Designed with Broadcom, Manufactured by TSMC：Meta plans to begin production of its latest custom AI chips in September, using a modular design approach in partnership with Broadcom and manufactured by TSMC. The company aims to reduce dependence on Nvidia GPUs while spending an estimated $125 $145 billion in capital expenditures this year. Source: techcrunch.com
Policy
- Google to label ads created or edited with generative AI：Google announced a new feature that will disclose when advertisers have used generative AI tools to create or edit their ads. The move adds transparency to one of the world's largest digital advertising platforms. Source: techcrunch.com
- How Did the US Government Decide OpenAI's Frontier Model Was Safe to Release?：TechCrunch published an investigative piece examining the opaque process behind the US government's safety approval for OpenAI's latest frontier model. The report reveals that details of the dialogue between the government and companies like Anthropic and OpenAI remain unclear, raising transparency concerns. Source: techcrunch.com
- Exclusive: Beijing considers curbing overseas access to China's top AI models — Reuters：Beijing is exploring measures to restrict overseas access to China's leading AI models, according to sources cited by Reuters. The potential policy shift could fundamentally reshape the global AI supply chain and open source ecosystem. Source: news.google.com
Agents
- OpenAI Shuts Down Atlas Browser, Moves Agent Features to Desktop App and Chrome Extension：OpenAI is sunsetting its AI powered browser Atlas after less than a year on the market. The company will instead transfer Atlas's agentic browsing capabilities to its desktop application and a Chrome extension. Source: techcrunch.com
- xAI Unveils Grok 4.5: Coding Agent Costs Slashed 80% with Near Frontier Speed：xAI has released Grok 4. 5, a new model that cuts coding agent operational costs by approximately 80% while maintaining near frontier inference speed. However, the model shows higher hallucination rates, suggesting accuracy trade offs in its design. Source: news.google.com
- Memetoro $MT Launches AI Agent Capable of Machine Payments via Stablecoins, No Bank Account Needed：Memetoro's $MT token has introduced an AI agent that uses stablecoins and the x402 protocol to conduct autonomous machine to machine payments without a traditional bank account. The project demonstrates a practical application of AI agents in decentralized finance. Source: news.google.com
Open Source
- World's first 'embodiment native' pretrained model LingBot VA 2.0 goes open source：Researchers have released LingBot VA 2. 0, claimed to be the world's first 'embodiment native' pretrained model, and made it fully open source. The model is designed to give robots a 'brain' that truly understands the physical world from the ground up. Source: qbitai.com