Anthropic focuses on 'Constitutional AI,' a method where models are trained to follow a specific set of rules to prevent harmful outputs.
Open the full topic