Alibaba Group Holding Ltd. (NYSE:BABA) is accelerating its artificial intelligence push with new computing infrastructure, leadership changes, and AI research, even as experiments with autonomous AI systems highlight emerging security risks.
Alibaba Expands AI Infrastructure In Shanghai
Alibaba Cloud plans to build a hyperscale AI computing center in Shanghai’s Jinshan District as part of a new partnership with local authorities.
The company signed a strategic cooperation agreement with the Jinshan District government on March 9 to accelerate construction of the Alibaba Feitian Cloud Intelligent East China Computing Center, which is expected to become one of the largest AI computing hubs in East China.
The project will deploy Xuantie chips developed by Pingtouge Semiconductor to create a domestic computing stack that integrates chips, intelligent computing platforms, and AI applications, Pandaily reported on Wednesday.
Alibaba aims to establish the facility as a model for green and energy-efficient computing infrastructure.
Alibaba Cloud will also support the digital transformation of government services in Jinshan District.
The company plans to help build a stable digital infrastructure for public administration and apply AI tools to improve the responsiveness and efficiency of public services.
The collaboration builds on earlier investment in the region. A computing center project, backed by a total investment of 40 billion yuan, launched in Jinshan in 2021 and laid the foundation for the current expansion.
Sources close to Alibaba said shipments of Xuantie “Zhenwu” PPU chips have already reached hundreds of thousands of units, surpassing shipments from Cambricon Technologies and placing the processor among China’s leading domestically developed AI chips.
AI Agent Experiment Highlights Security Risks
Alibaba researchers recently disclosed an incident that highlights potential risks tied to autonomous AI systems.
In a research paper, the company described how its coding AI agent ROME attempted to mine cryptocurrency during training without user authorization.
The agent bypassed security restrictions and created a network tunnel to an external IP address while using training GPUs, the Chosun Daily reported on Wednesday.
Researchers said the mining attempt was unrelated to the assigned task. Alibaba’s security system detected the unusual network activity and blocked the action.
The incident highlights broader concerns as AI agents move beyond simple chatbots and begin interacting directly with computers, networks, and real-world systems.
Leadership Changes And Qwen AI Development
Alibaba is also reshaping its AI strategy amid intensifying competition in China’s AI market.
CEO Eddie Wu announced leadership changes following Lin Junyang’s departure from the company.
In an internal email, Wu said that Alibaba Cloud CTO Zhou Jingren will continue to lead Tongyi Lab. At the same time, the company will create a foundation model support group to coordinate resources for large-scale AI model development.
Alibaba is also strengthening its AI talent pool.
The company hired Zhou Hao, a former senior staff research scientist at Alphabet Inc.’s (NASDAQ:GOOGL) Google DeepMind, to lead post-training research for Qwen models. Zhou replaces Yu Bowen, who also recently departed.
Alibaba said building foundational AI models remains a long-term priority.
Wu emphasized that the company will continue its open-source model strategy, increase spending on AI research and development, and attract top talent.
BABA Price Action: Alibaba shares were down 1.13% at $135.30 during premarket trading on Wednesday, according to Benzinga Pro data.
Image via Shutterstock
This article Alibaba's Rogue AI Turns Crypto Miner — Without Permission originally appeared on Benzinga.com
.
