All published AI news and analysis.

Proaction leverages OpenAI's Codex, GPT-Live-1, and GPT-6 Astra to accelerate the development, operation, and sales of modern fleet management solutions. This integration has reportedly led to significant improvements in sales and operational efficiency.

Researchers have introduced Ego-Exo4D-HM, a large-scale dataset featuring 4D human motion reconstructions derived from the Ego-Exo4D dataset's synchronized egocentric and multi-view exocentric video captures. This release also includes the accompanying reconstruction pipeline, which adapts state-of-the-art methods to leverage the multi-camera setup for improved accuracy.

OpenAI has released version 0.157.0 of its Codex Command Line Interface (CLI). This update focuses on tooling for developers interacting with OpenAI's Codex models.

LiquidAI has introduced LFM2.5-VL-DSpark, a new dataset designed to accelerate the training of vision-language models. The dataset aims to improve the efficiency and performance of these models.

Researchers have developed the Benchmark for Retrieving Information in EHRs (BRIE), a scalable framework that automatically generates question-answer pairs from longitudinal EHR notes. This "living" benchmark aims to provide continuous, up-to-date evaluation of clinical LLMs, addressing limitations of static, manually curated datasets.

A new study evaluates coding agents, including Claude Code (Opus 5) and Codex (GPT-5.6 Sol, GPT-6 Astra), on generalized Task and Motion Planning (TAMP) problems. The agents demonstrated effectiveness by synthesizing programs that generalize across problem instances, outperforming traditional hand-engineered planners and LLM-based baselines in success rates.

Google DeepMind has introduced Gemini 3.8 Live with Live Avatar, integrating real-time visual presence into its conversational AI for enterprise use. This feature enables dynamic visual personas with precise lip-syncing and natural expressions, enhancing customer service and interactive experiences.

Harvey, a legal AI platform, is now leveraging GPT-6 Astra to generate more structured and context-aware legal documents. This advancement aims to enhance efficiency for legal professionals.

Researchers have developed a novel method to optimize data annotation strategies for off-policy evaluation in offline reinforcement learning. The approach focuses on maximizing the efficiency of limited annotation budgets, particularly when dealing with complex, unstructured data like text or images.

Google's Private AI Compute platform is introducing a new private, server-side memory layer. This feature aims to provide persistent, cross-device AI memory while maintaining on-device privacy standards.