AI News

Follow the latest AI and automation news

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Mira Murati Wants Her AI to ‘Keep Humans in the Loop’

Mira Murati, founder of Thinking Machines Lab and former CTO at OpenAI, articulated her vision for AI development, emphasizing that her goal is not to replace humans but to create AI systems that enhance collaboration between humans and machines. This approach aligns with current trends in AI that prioritize partnership over replacement. This vision is particularly relevant amid rising concerns about AI's impact on the job market. Studies indicate that many jobs may be affected by new technologies, increasing the need for new business models.

Infrastructure

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

SMIC: AI Boom Drives Power Management Chip Demand, Automotive BCD Platform in High Demand

SMIC's CEO, Zhao Haijun, announced during an earnings call that the AI boom is significantly increasing demand for power management and data transmission chips. Meanwhile, supply chains for NOR Flash and other memory products are facing constraints. The company’s standalone Flash and analog process platforms are experiencing robust demand, showcasing their refined capabilities. SMIC's automotive-grade processes cover a wide range of products, indicating ongoing advancements in this sector.

Models

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

DeepMotor: Beijing Startup Betting on First-Person Human Data for Embodied AGI

Beijing-based startup DeepMotor, founded by Chen Kai, is advancing towards achieving embodied artificial intelligence through human video data. Established in early 2025, the company faced skepticism from local investors regarding the viability of this approach. However, the landscape shifted in May 2025 when Tesla announced its pivot to training robots using human video data. In June, GeneralistAI showcased a traditional learning model for robots, emphasizing the necessity of real human data. FigureAI also announced partnerships with commercial real estate firms to gather human video data.

💡

Tencent Q1 2026: Revenue Up 9% as AI Investment Surges

What is the impact of Tencent's AI investments on its profitability?

Agents

Former Kepler Robotics CEO Launches New Venture Sota Unlimited, Betting on 'Robot Brain' for Overseas Markets

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Agents

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Former Kepler Robotics CEO Launches New Venture Sota Unlimited, Betting on 'Robot Brain' for Overseas Markets

On May 14, Hu Debo, former CEO of Kepler Robotics, launched his new venture, Sota Unlimited, focusing on embodied AI. Unlike Kepler, which targeted industrial applications, Sota aims to develop the AI 'brain' for robots, tackling challenges in understanding motion and physics. Kepler Robotics, co-founded in 2023, quickly became a leader in bipedal humanoid robots, raising a billion-yuan A++ round in April. Sota Unlimited seeks to address complex issues in integrating robots into the physical world through advanced data collection and world action models.

Business

NTT DATA research reveals growing privacy and sovereignty barriers

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Business

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

NTT DATA research reveals growing privacy and sovereignty barriers

NTT DATA, led by Abhijit Dubey, has released new research indicating that enterprise AI is outpacing the underlying infrastructure. The report highlights a widening gap between organizations redesigning AI for privacy and sovereignty and those using unsuitable environments. Over 95% of respondents recognize the importance of private and sovereign AI, yet only 29% prioritize it concretely. Additionally, 35% of executives face challenges in building complex AI models within private or sovereign settings.

Models

Baidu steps up AI offensive with new model committee

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Models

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Baidu steps up AI offensive with new model committee

Baidu has announced the establishment of the Baidu Model Committee (BMC) to enhance its AI development efforts. This committee aims to centralize decision-making for R&D, technology planning, and application implementation in AI. This initiative reflects Baidu's strategy to strengthen its position in the AI market by improving coordination among different teams and achieving better outcomes in AI projects. The BMC is expected to streamline processes and foster innovation.

Agents

Alibaba launches Qoder 1.0 to automate software development with AI agents

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Agents

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Alibaba launches Qoder 1.0 to automate software development with AI agents

Alibaba has launched Qoder 1.0, a tool designed to enhance multitasking across projects and provide a knowledge-sharing engine at the team level. This tool aims to improve the efficiency of development teams by utilizing AI agents. Qoder 1.0 accelerates software development processes by automating repetitive tasks, allowing developers to focus on more complex project aspects. The tool features new functionalities that facilitate collaboration and knowledge exchange among teams.

Governance

Anthropic Proposes $1.5 Billion Settlement in Copyright Lawsuit

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Governance

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Anthropic Proposes $1.5 Billion Settlement in Copyright Lawsuit

A federal judge in San Francisco reviewed a proposed $1.5 billion settlement from Anthropic to resolve a lawsuit filed by authors who accused the company of using their books without permission to train the chatbot Claude. Despite the substantial amount, Judge Araceli Martinez-Olguin did not grant final approval during the session, requesting more information regarding the distribution of funds. This case is among dozens filed by copyright holders against tech giants concerning the training of large language models.

Models

ACL 2026: Alibaba DAMO Academy's I2B-LPO Breaks RLVR Homogenization — From Repetitive Sampling to Effective Exploration

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

Models

15 May 202628 Dhuʻl-Qiʻdah 1447 AH

ACL 2026: Alibaba DAMO Academy's I2B-LPO Breaks RLVR Homogenization — From Repetitive Sampling to Effective Exploration

The I2B-LPO framework has been accepted at the ACL 2026 conference, aiming to enhance exploration strategies for reinforcement learning models post-training. By improving exploration behavior, the framework achieves an increase in model accuracy of up to 5.3% and a semantic diversity boost of 7.4% across various mathematical metrics. Reinforcement learning models with verifiable rewards (RLVR) are modern approaches that enhance model capabilities in mathematics and coding. These models utilize multiple thought pathways for the same problem, reinforcing correct paths while minimizing errors.