All AI News and Updates of September 22–28, 2025

During the week of September 22–28, 2025, artificial intelligence advanced in several directions at once: from MIT’s new tool for material discovery to Nvidia’s $100 billion investment in OpenAI. Meanwhile, governments in China, Singapore, and South Korea strengthened regulations and training, while Europe focused on digital sovereignty.

Here’s a chronological overview of the key news.

 

📅 September 22, 2025

🧪 SCIGEN — AI for Breakthroughs in Materials Science

MIT researchers introduced SCIGEN, a tool that allows the introduction of constraints into generative models. This helps create materials with predefined exotic properties, such as for quantum systems.
🔗 Source: MIT News

 

💰 Nvidia to Invest $100 Billion in OpenAI

Nvidia and OpenAI signed a letter of intent for strategic cooperation. Nvidia will provide infrastructure and chips, while OpenAI gains resources for scaling. This marks the largest step in the history of AI partnerships.
🔗 Source: AI News

 

🏛️ Meta’s Llama Approved for U.S. Federal Agencies

The U.S. General Services Administration (GSA) approved the use of Llama in federal agencies. This is the first time an open-source model has been authorized for government use.
🔗 Source: crescendo.ai

 

🌍 UN: AI and Climate

The UN climate chief stated that AI is already helping optimize energy systems and reduce emissions. But with the rising energy consumption of data centers, stricter regulations and balance between benefits and risks are required.
🔗 Source: The Guardian

 

📅 September 23, 2025

🏠 Tado — Smart Heating with AI Assist

European company Tado launched the AI Assist update for its smart thermostats: adaptive heating, return-home predictions, and detailed energy analytics.
🔗 Source: The Verge

 

📅 September 24, 2025

🔒 Cisco Launches Sovereign Critical Infrastructure in Europe

Cisco announced a portfolio of solutions for sovereign infrastructure: local data centers, air-gapped systems, and compliance with European security standards.
🔗 Source: Cisco Newsroom

 

☁️ Alibaba Cloud Expands with Data Centers in France and the Netherlands

Chinese cloud giant Alibaba announced expansion in Europe with new data centers. This move strengthens the presence of Asian providers in the European AI and cloud market.
🔗 Source: European Business Magazine

 

📅 September 25, 2025

🔧 TSMC and Qualcomm — New Directions in AI Chips

TSMC is implementing AI in chip design, while Qualcomm introduced processors with remote control features, optimized for AI workloads on PCs.
🔗 Source: TechWire Asia

 

🇨🇳 China: Over 1,500 Large AI Models Registered

Chinese media reported that more than 1,500 AI models had been registered by September, confirming the explosive growth of the country’s AI industry.
🔗 Source: news.cn

 

📚 Singapore: Mandatory AI Training for Civil Servants

The government announced that all civil servants must undergo training in AI literacy to effectively use AI tools at work.

 

📖 China Publishes White Paper on AI Agent Safety

A consortium of academic and industrial organizations released a White Paper on AI Agent Safety Governance, outlining risks and regulatory strategies.

 

📅 September 22, 2025 (South Korea)

🇰🇷 South Korea: Five-Year AI Development Strategy

The South Korean government included AI development in its new five-year economic plan, aiming to place the country among the world’s top three AI leaders.

 

📊 Conclusion

The week of September 22–28, 2025 highlighted several global AI trends:

• Infrastructure and regulation take center stage. Nvidia and OpenAI sealed a record partnership, Cisco and Alibaba expanded capacity, and governments advanced rules and training programs.
• Science moves forward. 🧪 MIT’s SCIGEN demonstrates how AI can drive material discoveries with exotic properties.
• AI enters everyday life. 🏠 From Tado’s smart thermostats to 📚 Singapore’s mass civil service training — AI is becoming deeply embedded in daily practices.
• Laying the foundation for the future. Many announcements focus not on ready-to-market products, but on creating conditions: infrastructure, standards, and education. These steps pave the way for the next generation of AI solutions we’ll see in the coming years.