Topic
Explore the latest artificial intelligence news and research.
Chiba University and collaborators have unveiled a method that generates event camera output directly from 3D space. Time search and statistical early termination cut computation time. We explain how it differs from prior work, what the "about one-third" comparison means, and the challenges that remain for real devices.
Korean media report that NVIDIA wants suppliers to speed up glass substrate development, with adoption possible before 2028. TSMC's test package and delays at Samsung Electro-Mechanics and SKC show that large-scale customer qualification is still unconfirmed.
Anthropic reports that Z.ai's open-weight GLM-5.3 comes close to Claude Mythos Preview on certain exploit-development tests. We cross-check primary sources from NIST and Z.ai, explain why same-name benchmark scores differ, note the limits of the simulated tests, and consider what it means for defenders.
The White House and six leading AI companies agreed to a four-tier voluntary safety oversight system. We look at the gap between board-level reporting and public disclosure, and how the accord relates to existing government evaluations and the executive order renaming AI as SI.
ASML CEO Christophe Fouquet expects AI's impact to unfold over 10 to 15 years. Company documents show the gap between expanding current EUV output and adopting High NA, plus supplier limits and AI use in chipmaking with Mistral AI.
OpenAI has announced the Decisions API in limited preview. It uses GPT-6 Luna to pick from set options for classification and next actions. We look at how it differs from Structured Outputs, the conditions behind the reported 150 ms figure, and how it compares with Jev.
Reuters reports on McDonald's AI pricing engine. Read against its 10-K and earnings call, the tool may matter less for guessing what customers will pay than for how a franchisor that collects about 14.2% of franchisee sales can shape pricing, and what could undermine that reading.
OpenAI's new $500-a-month Pro 500 adds Astra Ultrafast. Here's what the "up to 8x" speed applies to, how it draws down plan limits (8x) and purchased credits (6x), and how Work and Codex share one allowance.
At DevDay, OpenAI introduced reusable development environments for Codex Cloud. We look at how they differ from the old setup, what state each task keeps, how network secrets avoid exposing API keys, and the permissions and limits of internal-network access.
OpenAI announced ChatGPT Space, a workspace where people and AI co-edit shared documents called Pages. Based on official materials, we look at how it differs from Projects, what recurring updates require, and how links and summaries change who can see your information.
