Anthropic

An AI safety and research company that makes the Claude AI models. Founded in 2021, it aims to develop advanced AI responsibly for the long-term benefit of humanity.

About Anthropic

Anthropic is an AI safety and research company that builds Claude, a family of AI models and assistants used by individuals, developers and businesses. Anthropic describes its aim as the responsible development and maintenance of advanced AI for the long-term benefit of humanity, and it puts as much weight on making AI reliable and safe as on making it capable.

The company was founded in January 2021 by seven former OpenAI employees, including the siblings Dario Amodei, now chief executive, and Daniela Amodei, now president. Its team also includes co-founders working on research, policy and interpretability, the study of what is going on inside an AI model. It is based in San Francisco and operates as a public benefit corporation, a type of company that is legally required to consider a public benefit as well as profit.

Safety is at the heart of Anthropic's mission. Its researchers work on making AI systems that are reliable, interpretable and steerable, which means people can understand them and guide their behaviour. A well-known idea from the company is Constitutional AI, a way of training a model to follow a written set of principles. Anthropic says it is one part of a larger effort and works with government, academia, non-profits and industry on safety. It lists values such as acting for the global good, weighing both the risks and the benefits of AI, and putting the mission first.

Claude launched in 2023 and has had several generations since. Anthropic offers Claude through apps, through Claude Code, which helps developers write software, and through a developer platform and partnerships with the major cloud providers. Much of its research is published, from interpretability studies to work on how AI models behave.

Anthropic describes its culture as high-trust and low-ego, with direct communication and an assumption of good intentions, and expects everyone to contribute whatever their role. Teams are interdisciplinary and its people come from backgrounds that include NASA, startups and the armed forces. It hires across research, policy, product and operations. The pages we read do not list office locations or describe a student scheme, so students should check the careers page for current roles and where they are based.

What they do

• {'desc': 'AI models and assistants for writing, analysis, coding and more.', 'title': 'Claude'}
• {'desc': 'A coding assistant that helps developers build and fix software.', 'title': 'Claude Code'}
• {'desc': 'API access so businesses can build on Claude.', 'title': 'Developer platform'}
• {'desc': 'Work on keeping AI systems reliable, honest and under human guidance.', 'title': 'Alignment and safety research'}
• {'desc': 'Studying what happens inside AI models so people can understand them.', 'title': 'Interpretability research'}

Key facts

• Founded: January 2021
• Headquarters: San Francisco, USA
• Founders include Dario Amodei (CEO) and Daniela Amodei (President)
• Structure: public benefit corporation
• Known for the Claude models and Constitutional AI

Website: www.anthropic.com