Password manager 1Password has introduced a new tool, 1Password for Agents, that allows users to grant autonomous AI agents access to their personal accounts without handing over their credentials or passwords to the AI.
Password manager 1Password has introduced a new tool, 1Password for Agents, that allows users to grant autonomous AI agents access to their personal accounts without handing over their credentials or passwords to the AI.
The new service solves one of the main security problems, as users increasingly trust personal AI assistants to perform routine tasks on the Internet, writes TechCrunch.
1Password for Agents builds a secure communication system between the AI agent, the user, and the password manager.
Authentication: When an AI agent attempts to log in to an account on behalf of a user, it sends a unique digital key to 1Password to verify its authenticity.
Confirmation: The user receives a notification about the login attempt and personally confirms the request.
Authorization: 1Password automatically fills in the necessary data and logs in for the assistant without revealing the passwords or access keys themselves.
“Handing over a password or secret key is the easiest way to work with an agent, but it leaves sensitive data in places you no longer control. 1Password for Agents makes it possible to securely delegate tasks to personal AI assistants while maintaining full control over your credentials,” the company noted.
1Password for Agents supports integration with a wide range of personal assistants, including:
Anthropic (Claude Duet);
OpenAI (Dots and ChatGPT Work);
xAI (Grok Bot);
Meta (Muse);
applications from the startup Town.
The company is already working on expanding the list of partner platforms.
The launch of the tool comes amid growing concerns among financial institutions about the activity of AI agents. According to Axios, US banks have begun massively recording automated actions inside customers' personal accounts.
This is not about hacking or theft of funds, but about situations when users themselves give assistants logins and passwords for managing finances. The main problem is that banking systems cannot clearly determine whether AI is fulfilling a confirmed request from the client or acting autonomously. Representatives of the banking sector call this situation alarming, and the banking holding Chase emphasizes that customers should not transfer their credentials to third parties or services under any circumstances.
