1Password has launched a new integration that lets Anthropic's Claude sign into websites and apps on your behalf without ever exposing your actual credentials. The feature, which 1Password calls Agentic Mode, is designed to solve one of the more awkward problems in the rise of AI agents: how a chatbot can complete real tasks that require a login without you handing over your passwords.
According to ZDNet, the integration can enter passwords and multi-factor authentication (MFA) codes without exposing those credentials to Anthropic or to the underlying model. Business Wire framed the announcement similarly, saying the partnership lets an AI agent use approved credentials without exposing the secret to the model.
The Verge reports that 1Password for Claude is a browser integration allowing the chatbot to access stored credentials such as usernames and passwords, so users can authorize Claude to complete multi-step tasks like booking travel and managing online accounts. Engadget describes it as a way to use Claude agents for personal chores without exposing your credentials.
According to 9to5Mac, as reported via Techmeme, the integration is launching for Mac users and lets Claude sign in to websites without seeing the user's password or 2FA code. Coverage from outlets including SiliconANGLE, IT Pro, and Crypto Briefing echoed the core pitch: secure credential access for an AI agent, with Crypto Briefing suggesting it sets a new standard for AI identity security.
The common thread across every source is the security model — the agent gets the access it needs to act, while the secret itself stays hidden from both the AI company and the model doing the work.
Why it matters: as AI assistants move from answering questions to actually doing things on the web, the question of how they handle your logins becomes central to whether people trust them — and this is an early attempt to let agents act on your behalf without asking you to give up your most sensitive secrets.