All published AI Radar articles.

KiwiFS is an open-source Markdown filesystem designed for AI agents and human teams, offering features like search, versioning, and structured data capabilities. It aims to bridge the gap between human-readable Markdown files and the structured data needs of AI agents.

GitHub Copilot has integrated Gemini models into its CLI, cloud agent, and Copilot application, enhancing its AI capabilities for developers.

GitHub Copilot's cloud agent now supports scheduling and automating tasks, enhancing its utility for developers.

GitHub Copilot has introduced a new feature, `/chronicle`, designed to provide developers with insights into their agent sessions.

OpenAI has announced new Codex plugins, sites, and annotations designed to enhance productivity for various professional roles, including analysts, marketers, designers, and investors, by integrating AI into their workflows.

Travelers has implemented an AI-powered Claim Assistant, developed in collaboration with OpenAI, to enhance customer support for filing claims and manage operational scaling during high-demand periods.

OpenAI is advocating for global collaboration to establish an AI Safety Institute specifically dedicated to addressing youth safety and opportunities in the context of artificial intelligence.

OpenAI's "The Next Era of Knowledge Work" report details how Codex is enhancing productivity across various knowledge work domains, including research, data analysis, workflow automation, and content creation.

Researchers have identified and analyzed "Perceptual Judgment Bias" in multimodal large language models (MLLMs) when used as evaluators. This bias causes MLLMs to prioritize plausible textual narratives over perceptually accurate visual evidence. To address this, they introduced the Perceptually Perturbed Judgment Dataset and a unified training framework combining GRPO-based rewards with a batch-ranking objective, aiming to improve perceptual fidelity and alignment with human evaluation.

Researchers have introduced Staged Executable Inverse Graphics (SEIG), an agentic framework that reconstructs 3D scenes from single images directly into editable Blender programs using pretrained vision-language models (VLMs). This method refines scene elements like geometry, materials, and lighting in Blender code, demonstrating improved reconstruction fidelity through task decomposition.