24-hour AI news at a glance: OpenAI upgrades real-time voice models, domestic large model financing heats up, and global AI regulation and computing power competition intensify
In the past 24 hours (May 7 to May 8, 2026), the core changes in the AI industry have focused on three main lines: cutting-edge models continue to expand to voice, cybersecurity, and agents; China's large models and AI infrastructure financing is active; European and American regulatory, military and corporate organizations are being reshaped by AI.
- OpenAI released a new generation of real-time voice models
OpenAI has launched three API audio models: GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper, covering real-time conversation, translation, and streaming transcription. The new model emphasizes understanding, translation and callable tools, and the voice agent is moving from a "chat entrance" to a "task execution entrance".
- China and the United States may launch formal AI discussions
It is reported that China and the United States are considering including artificial intelligence in the topic of subsequent high-level meetings. If relevant mechanisms are implemented, AI security, computing power limitations, model risks and technology competition may become new priorities for scientific and technological communication between the two sides.
- The dark side of the moon completed about $2 billion in financing
The valuation of the dark side of the moon exceeded $20 billion after a new round of financing, with the participation of Meituan Dragon Ball and other institutions. Capital continues to concentrate on leading large model companies, indicating that domestic model competition is shifting from simply releasing models to productization, ecology and listing expectations.
- DeepSeek is said to have promoted the first round of external financing
Multiple media outlets said that DeepSeek may make its first external financing, with a valuation range of about $45 billion to $50 billion. If the financing is successful, it will mean that China's leading open source large model companies will enter a stage of higher competition in capital, talent and computing power.
- Anthropic received computing power support from SpaceX
Anthropic has reached a computing power cooperation with SpaceX and will use large-scale AI data center resources. The competition for computing power by cutting-edge model companies continues to escalate, and computing power cooperation is becoming another main front in addition to model competition.
- Wuwen Xinqiong completed more than 700 million yuan in financing
Wuwen Xinqiong announced that it has received more than 700 million yuan in financing, which will be used for multi-heterogeneous computing power, software and hardware collaboration and enterprise-level agent services. AI infrastructure companies are shifting from "providing computing power" to "improving the efficiency of token-to-productivity conversion".
- Sichuan released the "Artificial Intelligence +" No. 1 innovation project plan
Sichuan proposes that the scale of the artificial intelligence industry will exceed 400 billion yuan by 2030, and promote the implementation of scientific research, manufacturing, cultural tourism, energy and other scenarios. Local AI policies are guiding from the direction into the stage of industrial indicators and application scenarios.
- The EU temporarily adjusts the pace of high-risk AI rules
Relevant EU agencies have reached a provisional agreement on the adjustment of AI rules, and the implementation of some high-risk AI system requirements has been delayed. Regulatory levels are looking for a balance between safety constraints, industrial innovation, and corporate compliance costs.
- Cloudflare announced AI-first restructuring and layoffs
Cloudflare plans to lay off about 20% of its workforce and emphasizes that the company will restructure its business around AI-first principles. AI is changing the way technology companies operate internally, and job replacement and organizational efficiency are also the focus of the industry.
- Scale AI expands Pentagon AI cooperation
Scale AI's cooperation with U.S. defense-related departments has been expanded to $500 million for data analysis and decision support. The application of AI in defense, intelligence, and high-risk decision-making scenarios continues to accelerate, leading to governance and security discussions.
Frequently Asked Questions (Q&A)
Q: What are the most important changes in the AI industry in the past 24 hours?
A: The most important thing is that AI continues to expand from text models to voice agents, cybersecurity, enterprise processes, and defense scenarios.
Q: What is the focus of domestic AI journalism?
A: The focus is on large model financing, AI infrastructure financing and the implementation of local "artificial intelligence +" policies, and the industrialization signal has been significantly enhanced.
Q: What is the significance of OpenAI's new voice model?
A: It lowers the threshold for real-time voice application development and may drive the upgrade of customer service, translation, conferencing, education, and in-car voice agents.
Q: What are the main risks in the AI industry?
A: Risks are concentrated in computing power concentration, employment substitution, misuse of high-risk scenarios, network security and regulatory lag.