Claude is becoming more agentic. Amanda Askell is thinking through what that means

# Claude Is Getting Smarter at Making Decisions on Its Own—And That Raises Big Questions About Ethics As Claude evolves from a chatbot you chat with into an AI agent that can actually manage tasks like your investments without constant human oversight, Anthropic researchers are racing to build in ethical guardrails that match real-world complexity. The challenge is teaching Claude to understand what *you* value rather than imposing its own moral judgments, which becomes crucial when an AI is making consequential decisions over extended periods rather than just answering questions. Anthropic is tackling this by expanding and refining its internal "constitution"—a set of principles like safety and helpfulness—to handle scenarios that didn't exist when AI was just a conversation partner.