1Password might just have a solution to the password security problem posed by AI agents

News
By published

Secure Agentic Autofill solves common security problems with AI agents

An abstract image of a lock against a digital background, denoting cybersecurity.
(Image Credit: TheDigitalArtist / Pixabay) (Image credit: Pixabay)
  • AI agents are being provided access to credentials in their workflows
  • This can cause credentials to be fed into LLMs without user knowledge
  • 1Password has created a secure solution to this problem

AI agents are becoming prolific across industries, providing employees with an easy way to automate tasks through designated workflows, but in order to function they often require credentials to access systems.

This requires human interaction to enter login details and passwords, slowing down both the AI agent and human workers. As a result, AI agents are being provided credentials through agentic browsers.

This has created a new problem has been created, especially with AI agents in headless agentic browsers, as credentials can be exposed into the underlying large-language model - but 1Password, one of the best password managers, thinks it may have come up with a solution.

Secure Agentic Autofill

1Password has developed Secure Agentic Autofill which allows AI agents to use credentials within their workflow without ever being able to see or handle the credentials themselves.

Instead, credentials are stored within the 1Password browser extension, and can be requested by an AI agent. The browser extension will fill the credentials on behalf of the AI agent, providing greater security within the workflow without impacting efficiency.

1Password BrowserBase collaboration for Secure Agentic Autofill

(Image credit: 1Password)

Secure Agentic Autofill uses a new protocol that uses an encrypted connection between the device and the browser extension using the Noise Framework. The password is requested by the AI agent, a human approves the credential usage, and the AI agent continues on its workflow.

1Password has partnered with Browserbase in order to deliver Secure Agentic Autofill through a new browser automation workflow UI. Just like the AI agent, the Browserbase connection has zero access to the credentials stored within the 1Password browser extension, and all requests require human approval.

1Password BrowserBase collaboration for Secure Agentic Autofill

(Image credit: 1Password)

Ultimately, Secure Agentic Autofill prevents credentials from being spread across agents, logs, and prompts, or distributed into areas that identity access management tools cannot see.

Additionally, by using 1Password, credentials can be better managed, added, and revoked without the fear that a password may have been leaked into the far corners of an LLM.

Introducing 1Password Secure Agentic Autofill for Browserbase - YouTube Introducing 1Password Secure Agentic Autofill for Browserbase - YouTube
Watch On

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

You might also like

TOPICS
Benedict Collins
Benedict Collins
Senior Writer, Security

Benedict has been writing about security issues for over 7 years, first focusing on geopolitics and international relations while at the University of Buckingham. During this time he studied BA Politics with Journalism, for which he received a second-class honours (upper division), then continuing his studies at a postgraduate level, achieving a distinction in MA Security, Intelligence and Diplomacy. Upon joining TechRadar Pro as a Staff Writer, Benedict transitioned his focus towards cybersecurity, exploring state-sponsored threat actors, malware, social engineering, and national security. Benedict is also an expert on B2B security products, including firewalls, antivirus, endpoint security, and password management.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.