Anthropic's New AI Agent Can Act on Its Own—But Asks Before Sending Money or Messages

Anthropic has launched Claude Duet, a new AI agent that works independently to manage tasks, as competition with Meta's Muse and OpenAI's Dots intensifies. Unlike standard chatbots that only respond to prompts, Duet can carry out multi-step tasks on its own. However, the company has built in a safety check: the agent must get user approval before sending messages, making payments, signing documents, or deleting information. The approval requirement aims to keep users in control as AI agents take on more responsibility. Anthropic's move signals a growing race among tech firms to build AI that acts, not just answers.