1Password lets AI agents log in without seeing your password
1Password has launched a tool that lets AI agents sign in to your accounts on your behalf without ever being shown the password. You approve each login yourself.
AI agents, which carry out tasks for you such as booking, ordering or checking accounts, often need to log in to websites. Until now, the easy way to allow that has been to hand over your username and password. On 22 October, 1Password announced a safer alternative, called 1Password for Agents.
Here is how it works. When an agent wants to sign in, it first proves who it is to 1Password. You then get an alert asking whether to allow the login. If you say yes, 1Password types the details in itself, so the agent can finish the job without ever seeing the password.
The company's argument is that giving passwords to an agent can leave copies of sensitive information lying around where you can no longer control them. TechCrunch lists agents that can use the tool, including Anthropic's Claude Duet, OpenAI's ChatGPT Work and Meta's Muse. The article doesn't give pricing or a release date.
The launch comes as banks in the US are reportedly starting to spot automated agents acting inside customer accounts, and are unsure whether the customer actually asked for it. One bank advised customers not to share their login details with anyone.
What this means for your business
If you or your staff are trying out AI agents, don't paste passwords or banking logins into them. Keep a rule that an agent never gets full access to accounts holding customer data or money, and use a password manager that can approve each login separately.
Source: TechCrunch, 22 October 2026. This is our own summary; read the original for the full story.