The Archive

Our goal is to facilitate the development of AI-powered cybersecurity capabilities for defenders through grants and other support.

We've trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”). In addition to boosting performance relative to outcome supervision, process supervision also has an important alignment benefit: it directly trains the model to produce a chain-of-thought that is endorsed by humans.

Our nonprofit organization, OpenAI, Inc., is launching a program to award ten $100,000 grants to fund experiments in setting up a democratic process for deciding what rules AI systems should follow, within the bounds defined by the law.

Anthropic closes $450M Series C funding round to expand product deployment and AI reliability research.

Now is a good time to start thinking about the governance of superintelligence—future AI systems dramatically more capable than even AGI.

The ChatGPT app syncs your conversations, supports voice input, and brings our latest model improvements to your fingertips.

Zoom invests in Anthropic and announces Claude API integration for video conferencing platform.

Anthropic releases Claude with 100K token context window, enabling longer document and code analysis.

Anthropic describes Constitutional AI method: training LLMs with explicit principles for safer, more controllable outputs.

We use GPT-4 to automatically write explanations for the behavior of neurons in large language models and to score those explanations. We release a dataset of these (imperfect) explanations and scores for every neuron in GPT-2.