Topics

Explore how merging Graph Machine Learning with Large Language Models creates Graph Foundation Models, enhancing our ability to analyze interconnected data for advanced applications in various domains.

Imperial and Thomson Reuters report a way to adapt an existing open AI model so organisations can own a capable system without training one from scratch.

OpenClaw, an open-source AI agent project, and Moltbook, a bot-only social network, have surged in popularity since late 2025, attracting commentary from AI leaders.

The same architecture first shown on graphics processor code optimization also finished every public level of a demanding benchmark that requires discovering rules through trial and error.

Internal embodiment, combining physical interaction and self-awareness of internal states, could improve safety and trustworthiness of future AI systems.

Public records show clear rises in reported software flaws and some mathematical progress while optimization advances remain steady.

An updated version of GLM shows clear gains on complex programming work and unexpected ability to locate weaknesses in software, with public release of its parameters planned after safety checks.

The ERA system developed by Google and Harvard researchers dramatically speeds up the creation of custom programs used in scientific research and outperforms human-written code.

A plug-and-play system called UniSpec makes AI responses faster across different computers and languages while keeping the answers exactly the same.