Anthropic reportedly acquired biotech startup Coefficient Bio in a $400 million deal, indicating a significant expansion of the AI company into the intersection of AI and biological research.
Researchers introduce DataFlex, a novel framework that enables dynamic, data-centric training of LLMs by adapting data selection and augmentation strategies based on model performance and data characteristics.
OpenAI has announced an executive reshuffle, moving COO Brad Lightcap to a new role overseeing 'special projects,' signaling a strategic shift in the company's high-priority initiatives.
A comprehensive survey paper explores the concept of latent space in AI, detailing its foundational principles, evolutionary trajectory, underlying mechanisms, capabilities, and future outlook across various AI domains.
A new research paper introduces a 'Generative World Renderer' capable of creating complex and realistic virtual worlds, potentially revolutionizing simulation, gaming, and virtual reality applications.
SKILL0 presents a new approach to agentic reinforcement learning that allows AI agents to internalize skills directly from in-context examples, significantly improving their adaptability and learning efficiency.
Anthropic is increasing its engagement in political activities by establishing a new Political Action Committee (PAC), reflecting a growing focus on influencing AI policy and regulation.
OpenAI has announced the acquisition of TBPN, a move expected to bolster its capabilities in an undisclosed strategic area of AI development.
The ASI-Evolve paper introduces a novel framework where AI systems are designed to autonomously accelerate the development and improvement of other AI models, hinting at a future of self-improving AI.
Salesforce has launched an enhanced Slackbot AI agent, intensifying its competition with Microsoft and Google in the rapidly evolving workplace AI market.
Google has announced that its AI-powered video creation tool, Google Vids, will now allow users to create, edit, and share videos for free, making advanced video production more accessible.
OpenAI provides an in-depth look into its methodology for developing the 'Model Spec,' outlining the principles and processes guiding the creation of its advanced AI models.
EgoSim introduces an egocentric world simulator designed to generate realistic embodied interactions, providing a crucial tool for training and evaluating AI agents in complex environments.
UniDriveVLA proposes a unified vision-language-action model that integrates understanding, perception, and action planning to enhance the capabilities of autonomous driving systems.
New research explores 'Steerable Visual Representations,' enabling more controllable and interpretable manipulation of visual data through AI models.