1Password lets AI agents use stored logins without seeing the passwords
Agents can complete multi-site tasks with approved credentials, while the models themselves never receive the underlying secrets.
1Password has launched 1Password for Agents, a system that lets AI assistants from several providers, including Meta's Muse, OpenAI's Dots and SpaceXAI's Grok Bot, use logins stored in a user's vault without the models ever receiving usernames, passwords or vault contents. Access is granted per session, limited to approved items and not carried over, and passwords and one-time MFA codes are injected through a channel 1Password manages so they stay outside the model. This follows the Claude integration 1Password introduced in July and lets agents complete multi-step, multi-site tasks such as trip planning without asking the user for credentials at each step. The article notes that agents can still make ordering or booking mistakes, though this caution is the outlet's own view rather than 1Password's. Pricing and availability have not been disclosed.
Why it matters
For people who give AI assistants access to their accounts, the main change is that the credential itself stays out of the model. Access is limited to specific approved items and lasts only for one session, so an agent does not keep standing access to a user's vault. Passwords and one-time codes are passed through a channel 1Password controls, which narrows where those secrets can leak. Pricing and availability remain undisclosed, so how widely people will be able to use this is still unclear.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.
Sources
- Engadget