“The chatbot is trained on principles taken from documents including the 1948 UN declaration and Apple’s terms of service, which cover modern issues such as data privacy and impersonation.”
– Dan Milmo
The owners of a new generative AI called Claude 2 say it’s “constitutional AI.” They’ve trained the model to have ethical aspirations, helping position their product as the most aligned with human goals.
The model was developed by Antropic, a San Franciso start-up populated by several ex-employees of OpenAI, the first company to release a generative AI language model.
Claude 2 is currently limited to the US and UK markets.