1Password has launched a browser integration for Claude that allows Anthropic's AI assistant to use stored credentials with user authorization. The feature is designed for workflows where Claude needs to sign in or interact with services on a user's behalf.
The integration matters because AI agents are moving from answering questions to taking actions in browsers. Password managers could become a key control point for deciding when an agent can access an account and how that access is logged.
It also raises the stakes for agent security design. As assistants gain access to credentials and web sessions, products need clear permission prompts, audit trails, and limits that users can understand.