The company
Anthropic is an AI company founded in 2021 by former OpenAI researchers, including Dario and Daniela Amodei. It builds the Claude family of models and presents safety research as central to its work.
Constitutional AI
In a 2022 paper, Anthropic described Constitutional AI: instead of relying only on people to label good and bad answers, the model is trained to critique and revise its own responses against a written set of principles — a "constitution". The aim is behaviour that is more consistent, and principles that are easier to inspect.
Responsible Scaling Policy
In 2023 Anthropic published a Responsible Scaling Policy. It defines AI Safety Levels, loosely modelled on biosafety levels: as models become more capable, stricter security and deployment safeguards must be in place before they are trained further or released.
Looking inside the model
Anthropic also works on interpretability — methods for understanding what happens inside a neural network, so that behaviour can be explained and checked rather than only observed from outside.
Why it matters for users
Whatever model you use, the same questions apply: what rules guide it, how is it tested, and where does your data go?
Tools for this
Asaiejadoo — everyday calculation and AI guides, part of the ZIBADIS network founded by Masoud Moghaddam in Tehran.
