Latest AI News

# Microsoft Faces Copyright Lawsuit from Authors Over AI Training Data A group of authors has sued Microsoft, claiming the company used nearly 200,000 pirated books to train its Megatron AI system without permission or compensation. The lawsuit alleges copyright infringement as Microsoft incorporated these works into AI training datasets without seeking approval from rights holders. This case joins similar lawsuits against other AI companies like OpenAI, Meta, and Anthropic, highlighting growing tensions between tech advancement and intellectual property rights. The lawsuit raises fundamental questions about fair use doctrine in AI development, potential financial impacts on creators, and the future regulatory landscape for AI training data. The outcome could significantly influence how copyright law applies to artificial intelligence development.

Italy's A2A utility company has partnered with French firm Qarnot on an innovative data centre in Brescia that recycles server heat into the city's district heating network. Using liquid cooling technology, the facility will provide heat for over 1,350 apartments and prevent approximately 3,500 tons of CO2 emissions annually. This pioneering project demonstrates circular resource usage and could potentially scale across Lombardy, where half of Italy's data centres are located. The initiative addresses the environmental impact of digital infrastructure while supporting A2A's €16 billion sustainability investment strategy through 2035.

Three Mile Island nuclear plant is being prepared for a possible restart in 2027, one year ahead of schedule. Constellation Energy plans to revive Unit 1 at a cost of $1.6 billion, which would generate 835 megawatts—enough to power all Philadelphia homes. Tech companies like Microsoft are showing interest in purchasing this carbon-free electricity for data centers. The plant, infamous for its 1979 partial meltdown, now represents the growing nuclear revival in the US alongside other facilities in Michigan and Iowa. This project could serve as a model for repurposing dormant nuclear facilities to meet increasing clean energy demands.

Zhipu AI, a Chinese AI startup backed by Beijing, is rapidly expanding its global influence by targeting emerging markets in Asia, Africa, and the Middle East. The company offers sovereign AI infrastructure and hardware to governments and enterprises, often partnering with Huawei. With over $1.4 billion in state investment, Zhipu AI forms part of China's strategy to build a self-sufficient AI ecosystem competing with Western technology. Though OpenAI maintains a technological lead in some areas, Zhipu AI stands as China's top domestic model and filed for IPO guidance in April 2025. Its expansion represents a significant shift in the global AI power balance between China and the United States.