Zhang Chaoyang: People Need Real-World Interaction More—AI Makes Content Feel “Plastic”
On August 7, Zhang Chaoyang, founder, board chairman, and CEO of Sohu, said in a media interview that real human interaction and authentic content are things AI can never replace. Getting information and talking to a real person are two completely different things—AI simply can’t substitute for human communication at this point. In this ocean of information, people need to step out and connect with others more than ever. In the age of AI, genuine face-to-face interaction becomes even more precious.
OpenAI’s First Consumer Device Looks Like a Doughnut, Priced Above $300
On August 7, reports emerged that OpenAI’s highly anticipated new device will sport a distinctive design with movable parts, giving it a more pronounced personality. The price tag could exceed $300. The product is essentially a smart speaker without a screen, shaped like a doughnut and roughly the size of a hockey puck. The design philosophy is to let users hold it in one hand and move around the house with it. OpenAI hopes this device, slated for a 2027 release, will break new ground. It’s positioned as an AI-first computer that helps users tackle all kinds of tasks. The eye-catching design is also expected to help it stand out in the current smart speaker market, which is largely dominated by Amazon and Alphabet’s Google.
Unitree’s Wang Xingxing: High-Performance General-Purpose Robots Are the Key to Solving Labor Shortages
On the afternoon of August 7, Wang Xingxing, chairman, general manager, and CTO of Unitree Robotics, said during an online roadshow that driven by the combined pressures of an aging population, rising labor costs, and the need to boost productivity and quality of life, high-performance general-purpose robots capable of assisting humans across various tasks have become a vital breakthrough for tackling labor shortages and efficiency bottlenecks. The market demand here is massive.
OpenAI Updates ChatGPT Experience: Free Users Get GPT-5.6 Luna with Unlimited Text Chat
On August 6, local time, OpenAI announced an update to the ChatGPT model experience, improving GPT-5.6 Sol’s factual accuracy, response focus, and consistency across scenarios, while also expanding free users’ access to GPT-5.6 Luna. OpenAI said Plus and Pro subscribers will get the upgraded GPT-5.6 Sol, which adjusts response depth based on question complexity and cuts down on unnecessary formatting. For free users, OpenAI has switched the default model to GPT-5.6 Luna and widened access, offering unlimited text chat. For questions that need deeper reasoning, free users can now hit a new “Think” button to tap into higher reasoning capabilities.
Stripe Reportedly in Talks to Acquire OpenRouter for $10B, Moving Into AI Infrastructure
On August 7, reports indicated that payments tech company Stripe is in exclusive negotiations to acquire OpenRouter, an AI model aggregation platform, at a valuation of around $10 billion. OpenRouter had previously drawn acquisition interest from several major tech firms, but Stripe has now moved into the exclusive discussion phase.
Australian AI Infrastructure Firm Firmus Raises $2B, Backed by Nvidia, Blackstone, and Others
On August 7, Australian AI infrastructure company Firmus announced it had completed a $2 billion strategic equity funding round. Investors include Nvidia, Coatue, funds managed by Blackstone and other Blackstone investment vehicles, as well as global trading and tech firm Jane Street. Firmus said the capital will accelerate the rollout of its “Southgate AI Factory” project across Australia and set the stage for future expansion into other Asia-Pacific markets.
Xiaohongshu: Accounts Operating with AI Virtual Personas Must Clearly Disclose It on Their Profile
On August 7, Xiaohongshu announcedan upgradeto its AI governance rules, extending the proactive disclosure requirement from “notes” to “accounts.”Accounts runwith AI virtual personas are asked to clearly state this on their profile page and enable the “AIVirtual Person” label.Xiaohongshu noted that proactive disclosure won’t affect traffic; quality creators who identify themselves first get early access to the platform’s new AI tools and features. If accounts fail to disclose and get flagged by the system, they’ll face penalties ranging from throttling to other tiered measures.
Alibaba Launches AI Voice Platform CosyVoice Studio
On August 7, Alibaba officially rolled out CosyVoice Studio, a one-stop AI voice productivity platform. Built on its proprietary speech model Qwen-Audio, the platform includes voice recording tool CosyFlow, voice agent CosyAgent, and audio content creation tool CosyCreative. Starting today, users can search for “CosyVoice” in app stores to download it, with free access for a limited time. CosyAgent and CosyCreative are currently being tested with enterprise users via a whitelist invitation system and will open up to everyone soon.
Disney Tests AI Search Features to Boost Content Discovery Through Natural Language
On August 6, Disney announced that its ESPN and Disney+ platforms are testing AI-powered search and content discovery features, helping users describe what they want in natural language to find sports info and entertainment content faster.
