Anthropic

Anthropic is an AI safety and research company headquartered in San Francisco, founded in 2021 by a group that included several former OpenAI researchers.

Its model family is called Claude, sold through a consumer app, a developer API and cloud marketplaces. The company publishes a large amount of its own alignment and interpretability research, and its Constitutional AI approach - training a model against a written set of principles - is one of the more widely discussed techniques in the field.

It is a recurring name here for model launches, enterprise deals and the policy debate around how frontier systems should be evaluated before release.