AI
Build a Domain-Specific Embedding Model in Under a Day
A new tutorial from NVIDIA and Hugging Face shows how to build a domain-specific embedding model in under a day. The tutorial uses the Transformers li...
AI
AI
How we monitor internal coding agents for misalignment
OpenAI monitors its internal coding agents for potential misalignment using a technique called chain-of-thought monitoring. This involves analyzing th...
AI
OpenAI to acquire Astral
OpenAI has announced plans to acquire Astral, a company that develops AI-powered coding tools. The acquisition is expected to accelerate the growth of...
AI
State of Open Source on Hugging Face: Spring 2026
Hugging Face has released its quarterly report on the state of open source in the Hugging Face ecosystem. The report highlights trends and insights fr...
AI
Measuring progress toward AGI: A cognitive framework
DeepMind has proposed a cognitive framework for measuring progress towards Artificial General Intelligence (AGI). The framework aims to assess how clo...
AI
Holotron-12B - High Throughput Computer Use Agent
HCompany has released a high-throughput computer use agent called Holotron-12B. The agent is designed to efficiently utilize computing resources, pote...
AI
Introducing GPT-5.4 mini and nano
OpenAI has released two new versions of its GPT-5.4 model: mini and nano. These smaller models are designed to be faster and more efficient than the o...
AI
OpenAI Japan announces Japan Teen Safety Blueprint to put teen safety first
OpenAI's Japanese division has unveiled a plan to prioritize teen safety when it comes to generative AI. The 'Japan Teen Safety Blueprint' aims to imp...
AI
Equipping workers with insights about compensation
A recent study has found that Americans are using ChatGPT to ask questions about compensation and earnings. It's estimated that nearly 3 million daily...
AI
Why Codex Security Doesn’t Include a SAST Report
OpenAI's Codex Security uses a unique approach to vulnerability detection. Unlike traditional Static Application Security Testing (SAST), it relies on...
AI
Designing AI agents to resist prompt injection
OpenAI's researchers have designed a method to prevent AI agents from being manipulated through 'prompt injection'. This technique involves tricking t...
AI
From model to agent: Equipping the Responses API with a computer environment
OpenAI has equipped its Responses API with a computer environment. This allows the API to run secure, scalable agents that can access files, tools, an...