Introducing 1Password for Agents to securely manage AI logins without disclosing passwords

1Password has announced a new platform, 1Password for Agents, designed to let personal AI agents sign into websites and services without ever holding the user's passwords. The company says the system works by prompting the user when an agent hits a login screen: the user reviews the requested credential and approves it inside 1Password, after which the password manager autofills the login on the destination site. Under that arrangement, according to the company, neither the password nor the associated one-time code is exposed to the agent itself.

The announcement lists launch integrations with several agent products, including Muse from Meta, Anthropic's Claude Duet, OpenAI ChatGPT Work, [dots], Grok Bot, and Town Townie. 1Password said additional integrations with Instinct, Browserbase, Browser Use, Dana, KERNEL, Perplexity, Ramp, Tomo, and Wajo are planned. The company frames the offering around centralized credential management, arguing that keeping logins in a single vault means a password change only has to be made once, with connected agents continuing to use the current credentials rather than requiring updates across multiple systems. It also pitches the platform to developers, inviting agent builders to integrate so that “you build the agent, 1Password secures the sign-in”.

The launch reflects a broader push by security vendors to define how autonomous software gets authorized to act on a user's behalf, a problem that has grown more pressing as agents move from demonstrations into everyday tasks. 1Password positions the product as applying to both consumer errands and workplace agents that build applications and development workflows, describing the goal as letting “your agent do the work without handing over your passwords”.