1Password has introduced 1Password for Agents, a feature that lets AI agents sign into your accounts without seeing or handling the passwords involved.
Sign-in requests still need your approval, so an agent does not get open access to every account you have connected. The feature targets tasks where an agent has to log into various services to finish a job.
1Password CEO Aashish Tripathi framed it as a response to how agents are being used. "Agents are starting to do real work on our behalf," he said in a press statement. "The old model where you hand over a credential and hope for the best doesn't work."
That old model is the real target. Giving an agent a raw password means the software holds a secret that could be leaked, logged, or misused. Avoiding that kind of exposure is much of the reason password managers exist, and this applies the same logic to a new kind of user.
It also reflects a wider shift. As agents move from drafting text to acting inside accounts, the security question changes from what a model says to what it can reach. A password manager is a natural place to put a gatekeeper for that access.
The open questions are practical. The announcement as reported does not say which agents the feature supports, what it costs, or how the approval prompts will work day to day, and those details will decide whether people put up with the extra step.