NVIDIA Open Sources Reasoning Model for Autonomous Driving at NeurIPS 2025

LTIMindtree and Armada have announced a collaboration aimed at scaling edge and sovereign AI deployments, focusing on helping enterprises maintain data sovereignty, privacy, and operational continuity while implementing AI technologies at scale.

The Indian IT sector is witnessing a significant challenge as artificial intelligence (AI) struggles to penetrate senior hiring processes, which require nuanced, unstructured conversations that AI currently cannot replicate. This limitation raises questions about the role of AI in leadership recruitment, where human judgment and interpersonal skills are paramount.

Six Indian states are integrating artificial intelligence (AI) into everyday policing, utilizing technologies such as facial recognition, chatbots, and surveillance systems. This initiative aims to enhance law enforcement efficiency but has resulted in mixed outcomes and raised significant concerns regarding privacy and civil liberties.

IBM has partnered with Karnataka Bank to develop an API platform aimed at enhancing service delivery efficiency. This collaboration is expected to leverage IBM's technological expertise to streamline banking operations and improve customer experiences.

HSBC has entered into a multi-year agreement with Mistral AI to enhance the integration of artificial intelligence across its global operations, aiming to leverage AI technologies for improved efficiency and innovation.

Indian e-commerce firm Meesho is expanding its vernacular AI voice agents as it prepares for its Initial Public Offering (IPO), aiming to enhance the shopping experience through AI-driven personalization. This initiative reflects a growing trend in the integration of artificial intelligence in retail.

A new quantization method called Four Over Six (4/6) has been introduced to enhance the NVFP4 quantization algorithm, which is crucial for large language models (LLMs). This method evaluates two potential scale factors for each block of values, addressing issues of performance degradation during inference and divergence during training that arise from quantization errors in floating-point formats.