1Password’s new tool lets AI agents log in without ever seeing your password

Your AI assistant can now book flights without knowing your password. 1Password rolled out Secure Agentic Autofill on Tuesday, solving a security problem that emerged as artificial intelligence (AI) agents began handling online tasks. The feature lets AI complete purchases and log into websites without ever seeing user credentials.

AI agents from ChatGPT, Claude, and Gemini need passwords to work on websites. Once they have those passwords, they might store them, leak them, or expose them to large language models.

“By partnering with Browserbase, we’re making it simple for teams to build and innovate with agentic AI while keeping credentials and sensitive data protected by default,” said David Faugno, 1Password’s CEO.

The solution works like a security checkpoint. An AI agent requests credentials when it needs to log in somewhere. 1Password identifies the right password but stops everything until a human approves.

Source: 1Password

Users authenticate through Touch ID, Face ID, or similar methods. After approval, 1Password injects credentials directly into the browser via end-to-end encryption, completing its task without ever seeing or handling them.

Developing the protocol, 1Password partnered with Browserbase, which builds browsers for AI agents. The companies see this as essential infrastructure for an AI-powered internet.

“As we move toward an AI-native internet, these agents need a trust layer for credential access,” said Paul Klein, Browserbase’s CEO. “Our partnership with 1Password creates exactly that, extending their proven security framework into the emerging world of agentic browser automation so organizations can scale AI confidently.”

Aside from passwords, the system also automatically fills in two-factor authentication codes. Smart credential selection prevents phishing attacks by choosing the right login for each website.

Companies receive detailed audit logs showing exactly when and where AI agents used credentials. IT teams can manage both human and AI access from one dashboard. They can revoke access instantly if needed.

The system could benefit businesses as AI agents now handle procurement, customer service, and research tasks that require multiple logins. Without proper security, these automated systems create risks that IT departments struggle to control.

Early access started Tuesday for 1Password Enterprise Password Manager customers using Browserbase. The companies haven’t announced when other platforms might adopt the protocol.

The real test comes next. Major AI providers like Google, Microsoft, and OpenAI haven’t committed to using this security standard. Without their support, Secure Agentic Autofill might remain a specialized tool rather than an industry standard.