AI 新闻日报 2025-08-19：全球与中国 AI 头条 Top 10

1. Nvidia Develops New, More Powerful AI Chip for China

Nvidia is developing a new AI chip for the Chinese market, tentatively named the B30A, which will be based on its latest Blackwell architecture. Sources familiar with the matter report that the new processor will be more powerful than the H20 model currently sold in the region, signaling Nvidia's strategy to maintain its market presence in China while navigating U.S. export controls. The company aims to ship samples to Chinese clients for testing as early as next month.

2. OpenAI Rolls Out Landmark GPT-5 Model in August

OpenAI has officially launched its highly anticipated GPT-5 model. A key feature is a "Unified System" that automatically determines whether a user's prompt requires a quick response or deeper, more complex reasoning, simplifying the user experience. The new model series also includes variants like GPT-5 Mini for budget-friendly automation and GPT-5 Nano for on-device applications, making advanced AI more accessible across different platforms.

3. New EU AI Act Rules for General-Purpose AI Take Effect

A major milestone in AI regulation has been reached as new obligations under the EU AI Act for providers of General-Purpose AI (GPAI) models came into force on August 2, 2025. These rules require providers of models like large language models to ensure transparency, maintain detailed technical documentation, and comply with copyright laws. The AI Office, a new body within the European Commission, is now operational to oversee the implementation and enforcement of these regulations.

4. UK's National Health Service Trials AI to Speed Up Patient Discharges

The NHS is piloting an AI tool at a London hospital trust designed to accelerate the patient discharge process. The platform automates the creation of discharge summaries by extracting key information from medical records, which can save doctors significant time on paperwork. Officials believe this technology will help free up hospital beds, reduce waiting times, and allow medical staff to focus more on patient care.

5. Arm Hires Top Amazon Executive to Lead In-House AI Chip Development

In a strategic move to expand beyond licensing its designs, Arm has hired Rami Sinno, Amazon's director of AI chips. This high-profile hire is a clear indication of Arm's ambition to develop its own processors, a move that could see it competing directly with major customers like Nvidia and Apple who rely on its architecture.

6. Chinese Chipmaker Cambricon Raises $560 Million to Challenge Nvidia

Chinese semiconductor firm Cambricon Technologies is set to raise nearly US$560 million to fund its AI chip development. The fundraising effort is bolstered by the Chinese government's push for technological self-sufficiency and its directive for data centers to use more domestically produced chips, positioning Cambricon as a key player in China's efforts to compete with AI hardware leaders like Nvidia.

7. Study Reveals Medical AI Chatbots Can Be Easily Manipulated

New research has highlighted significant safety risks in relying on AI for health advice. A study published in the *Annals of Internal Medicine* found that well-known medical chatbots can be easily manipulated into providing authoritative-sounding but false health information. The findings raise alarms about the need for robust safety guardrails and user awareness as AI becomes more integrated into healthcare.

8. Indian Workforce Embraces AI to Future-Proof Careers, Report Finds

A new report from the job platform Indeed reveals that a large percentage of workers in India, including both white-collar and blue-collar professionals, are actively embracing AI. The survey of over 3,000 workers showed that 43% feel confident about using future technologies like Generative and Agentic AI, viewing them as essential tools for career growth and job security.

9. SoftBank Makes $2 Billion Bet on Intel Amid AI Push

Japanese investment giant SoftBank has taken a $2 billion stake in U.S. chipmaker Intel. This move is seen as a vote of confidence in Intel's turnaround efforts and aligns with SoftBank's massive investments across the AI ecosystem, which also include a recent $30 billion commitment to OpenAI.

10. Japan's KDDI Launches AI and Robotics Service for Office Design

Japanese telecommunications company KDDI has launched a new service that uses AI and robotics to design and operate smart offices, factories, and other commercial buildings. The service aims to optimize building layouts by simulating electrical wiring needs and robot traffic flows during the planning stage, which is expected to reduce construction costs and improve efficiency.