Anthropic details Claude Duet controls that keep approvals outside the model

Anthropic says Claude Duet keeps approvals and credentials outside the model in a US research preview; a beta user reports glitches.

Anthropic published the security and privacy design of Claude Duet, a proactive agent that monitors connected accounts and completes tasks in a cloud browser. Approval cards and an automated intent check sit outside the model so injected web content cannot answer them, and payments or account creation require approval each time. Credentials stay out of the model's reach, a sensitive-data classifier screens outgoing messages, and red teams plus an external penetration-testing firm evaluated the service. The app is in research preview for US Pro and Max subscribers. An unaffiliated beta user says Duet is useful but reports invisible threads, failed repository attachments, and weaker voice than Dot.