We focus on two areas: how Claude handles conversations about suicide and self-harm, and how we’ve reduced “sycophancy”—the tendency of some AI models …
Read more: Protecting the well-being of our users – Anthropic
Read more: Protecting the well-being of our users – Anthropic