The password manager checks the agent's identity with a digital key, asks for the user's consent, and then fills in the identifiers himself. Claude Duet, ChatGPT Work and Muse are already compatible.
On October 22, 1Password launched "1Password for Agents," a service designed for assistants who navigate and act on behalf of their users. Rather than giving the agent a password, a common but risky practice, the system establishes a channel between the agent, the person, and the password manager.
How it works
When an agent tries to connect to a site, it sends 1Password a unique numerical key that attests to its identity. The user receives an alert and must validate the request themselves; 1Password then enters the credentials in place of the agent, who never has access to them. The secrets are therefore not stored in the history or memory of a third-party assistant.
At launch, the service works with Anthropic's Claude Duet, OpenAI's Dots and ChatGPT Work, Meta's Grok Bot, Muse, and Town AI's agents, and the company says it is working to expand it. The announcement comes as U.S. banks are seeing an increase in agent activity in their customers' accounts.
Read the full article on the original source
Go to the source →