The Archive

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.

Anthropic partners with Scale AI to integrate Claude into enterprise data labeling and AI infrastructure workflows.

ChatGPT users can now turn off chat history, allowing you to choose which conversations can be used to train our models.

Anthropic advocates for NIST AI risk management framework and increased federal investment in AI safety standards.

This initiative is essential to our commitment to develop safe and advanced AI. As we create technology and services that are secure, reliable, and trustworthy, we need your help.

Ensuring that AI systems are built, deployed, and used safely is critical to our mission.

An update on our findings, the actions we’ve taken, and technical details of the bug.

We’ve implemented initial support for plugins in ChatGPT. Plugins are tools designed specifically for language models with safety as a core principle, and help ChatGPT access up-to-date information, run computations, or use third-party services.