The UK AI Security Institute and EvalEval are deepening their partnership by publishing verified, reproducible AI evaluation results through EvalEval's Evaluation Cards platform. AISI released benchma…
Universe is a newly launched software platform designed to measure and train AI general intelligence. It leverages the world's vast collection of games, websites, and other applications as environment…
RoboSchool is a free, open-source robotics simulation platform that works seamlessly with OpenAI Gym. It enables researchers, students, and hobbyists to experiment with robot control and reinforcement…
OpenAI formed OpenAI LP, a capped-profit company that lets it raise large-scale capital for computing and talent while capping investor returns. Excess profits flow back to the parent nonprofit, which…
MuseNet is a deep neural network that generates musical pieces up to four minutes long using up to ten instruments, spanning genres from country to Mozart to the Beatles. Trained on hundreds of thousa…
This article explains how the same transformer architecture used for text can generate images by treating pixel grids as sequences. Trained to predict the next pixel without labels, the model produces…
OpenAI's DALL·E 2 research preview reaches a milestone with over 3 million images generated by early participants, whose feedback has helped strengthen safety measures. The company will now invite up …
When releasing DALL·E 2, OpenAI faced the challenge that its text-to-image power could produce harmful content. The team implemented layered guardrails that filter both user requests and model outputs…
During the DALL·E 2 research preview, over 3,000 artists from 118+ countries integrated the AI image generator into their creative workflows. These early users acted as design partners, uncovering new…
DALL·E beta is now live, with OpenAI inviting 1 million users from its waitlist in the coming weeks. Creators can generate images using free credits refreshed monthly, and purchase additional credits …
A new embedding model has been released with three key improvements: significantly better capability in understanding nuance and context, greater cost effectiveness for high-volume workloads, and a si…
OpenAI has launched a text classifier that distinguishes AI-generated writing from human-written text, aimed at teachers, editors, and moderators. The article explains how the tool works by analyzing …
OpenAI has launched ChatGPT Plus, a pilot subscription program for its conversational AI assistant. Rather than a full-scale launch, this limited release lets OpenAI gather subscriber feedback and ref…
AI writing assistants handle three core tasks: generating fresh drafts, editing existing text, and iterating through rounds of feedback. They excel at creative projects like songwriting and screenplay…
GPT-4 is OpenAI's most advanced large multimodal model, accepting both images and text as inputs while producing text-only outputs. It achieves human-level performance on many professional and academi…
The ChatGPT iOS app offers more than the web version: conversations sync automatically across all devices, voice input enables natural hands-free interaction while cooking, driving, or brainstorming, …
OpenAI now supports fine-tuning for GPT-3.5 Turbo, allowing developers to train the model on custom input-output examples to achieve consistent tone, reliable structured outputs, and improved performa…
OpenAI is rolling out DALL·E 3 to ChatGPT Plus and Enterprise users. The release is backed by a multi-layer safety mitigation stack designed to reduce potential harms as access to the image generation…
OpenAI is rolling out a memory feature for ChatGPT, enabling the AI to retain details from past conversations—like your preferences, projects, and personal context—so future chats require less repetit…
This article explores the OpenAI mission: ensuring artificial general intelligence benefits all of humanity. It explains how the mission guides daily decisions, from prioritizing safety testing and al…