In the past 24 hours (May 5 to May 6, 2026), the main line of the AI industry has focused on three aspects: model iteration, computing power infrastructure, and regulatory review. In China, it focuses on end-side intelligence, industry agents and real-world scenario applications, while overseas it focuses on cutting-edge model security testing, cloud computing power orders and AI server requirements.
- OpenAI upgrades the default model of ChatGPT
OpenAI upgraded ChatGPT's default model to GPT-5.5 Instant, focusing on more accuracy, conciseness, and personalization. Its internal assessment said that in high-risk issues such as medical, legal, and financial, hallucinatory claims have decreased by 52.5% compared to the previous generation, indicating that the universal assistant is continuing to evolve towards a "low error rate".
- The United States expands security testing of cutting-edge AI models
The U.S. government has expanded the scope of risk assessment of unreleased AI models to include Google DeepMind, xAI and Microsoft. The focus of the test includes cyberattacks, chemical and biological weapons risks, and training data security, demonstrating that the regulation of cutting-edge models is moving from the Principles initiative to the technical review phase.
- Anthropic reportedly increased its purchases of Google Cloud and chips
Anthropic has reportedly committed to investing about $200 billion in Google Cloud and chip-related services over the next five years. This action reflects the lock-in demand of leading model companies for long-term computing power, and also strengthens the bargaining position of cloud vendors in the AI industry chain.
- AMD raised its performance forecast driven by AI demand
AMD expects second-quarter revenue to be higher than market expectations, mainly driven by demand for data center chips and AI infrastructure. The company's data center business revenue increased sharply year-on-year in the first quarter, indicating that the demand for AI inference and training is still supporting the semiconductor boom.
- Apple has reached a settlement over delays in Siri AI features
Apple agreed to settle a shareholder lawsuit stemming from delays in Siri's AI upgrade for $250 million. The incident shows that consumer electronics companies are facing higher cash pressure when promoting AI capabilities, and AI function delivery has become a focus of the capital market.
- China Mobile will launch AI-eSIM products
China Mobile announced that it will launch AI-eSIM at the 2026 Mobile Cloud Conference, which can schedule cloud models in real time, allowing terminals to think independently and respond instantly. The product is aimed at AI toys, smart wearables and other scenarios, reflecting the upgrade of the communication entrance to the end-cloud intelligent entrance.
- The two departments promote the "modulus resonance" action
The Ministry of Industry and Information Technology and the National Data Administration will promote the "analog-digital resonance" action in 2026, requiring the construction of high-quality data sets, industry models and characteristic agents around key industries. This action binds data, models and scenarios, and is expected to accelerate the implementation of AI in industrial, medical, automotive and other fields.
- AI accelerates into the "May Day" travel decision
During the "May Day" period, Fliggy and Xiaohongshu reported that the number of user consultations using AI increased by 56% compared with the Qingming holiday; The number of Meituan AI butler queries increased by 89% compared with the Spring Festival. Travel planning is becoming a high-frequency application scenario of AI assistants, and niche destination discovery and itinerary combination are the main needs.
- China's "artificial intelligence +" scenario continues to expand
According to Xinhua News Agency's survey, AI is entering the real economy such as Internet of Vehicles management, port scheduling, and driverless trucks. The scale of China's core artificial intelligence industry has exceeded 1.2 trillion yuan, and the penetration rate of AI applications in manufacturing enterprises above designated size has exceeded 30%, and industrial applications are moving from pilot to large-scale.
- The number of calls to China's AI large models has rebounded
The latest statistics show that the weekly call volume of China's large AI model rose to 7.942 trillion tokens last week, an increase of 81.7% month-on-month, once again surpassing the number of models on the list in the United States. The change in call volume shows that low-cost models and application ecosystems are promoting the expansion of domestic AI use.
Frequently Asked Questions (Q&A)
Q: What are the most important trends in the AI industry in the past 24 hours?
A: Model upgrades, computing power lock-in, and regulatory testing are advancing simultaneously, and AI competition has shifted from a single point of release to long-term infrastructure and trusted governance.
Q: What is the focus of AI news in China?
A: The focus is on industrial implementation, including end-side intelligence, industry models, cultural tourism consumption and manufacturing applications, and the core keyword is "scenario-based".
Q: What is the focus of overseas AI news?
A: The focus is on cutting-edge model security, cloud computing power procurement and chip demand, indicating that leading companies are building long-term barriers around computing power and compliance.
Q: What are the enlightenments for entrepreneurs and developers?
A: Opportunities are shifting from general-purpose chat products to vertical scenarios, agent workflows, and device-cloud combined devices, and AI products that can solve real business problems are more likely to form value.
