Wednesday, October 7, 2026

Meta to Create Internet Access Passes for AI Agents [Global AI Briefing]

Input
2026-10-07 08:46:18
Updated
2026-10-07 08:46:18
[Financial News] Meta has decided to create internet access passes that AI agents will need to carry out tasks such as shopping and booking flights. The move follows Amazon's blocking of Muse, the shopping AI agent Meta unveiled last month, from accessing its site. Meta says it plans to establish access rules that will allow websites to verify whose instructions an AI agent is following.
On the 6th (local time), Meta officially announced that it would develop a new internet standard, the Personal Agent Protocol, with Sierra, a U.S. startup that develops enterprise AI agents, to allow AI agents to operate on corporate websites.

On the 6th (local time), Meta officially announced that it would develop a new internet standard, the Personal Agent Protocol, with Sierra, a U.S. startup that develops enterprise AI agents, to allow AI agents to operate on corporate websites. The photo shows Meta CEO Mark Zuckerberg unveiling Muse, an AI agent for shopping. (Source: Yonhap News)

Companies in e-commerce, customer service and platform businesses—including Walmart, Stripe, Shopify, Genesys, Instinct, Rocket and NiCE—will participate in the development. AI agents need to access their services to shop and make payments.
Meta and Sierra said the Personal Agent Protocol is an open standard that will allow platforms to verify and manage whom an AI agent represents, what permissions it has been granted and what actions it is taking when accessing a company's website or API.
The authentication system will be designed to distinguish whether a user has given an AI agent permission only to look things up or also to carry out actions such as making actual purchases or reservations.
As AI agents such as Meta's Muse and OpenAI's Dot have increasingly come to market, browsing websites like people to search for products and even make purchases and payments on their behalf, cases of conflict with the security systems of major platforms in sectors such as shopping, air travel and healthcare have emerged. Platforms designed on the assumption that people would search for products and pay for them have begun identifying AI agents as abnormal traffic and blocking them.
Last month, Amazon blocked Meta's Muse from accessing its site. A legal dispute has also arisen over site access involving Perplexity's AI agent. Walmart's site has also had cases in which its existing verification procedures, designed to check whether a user is human, blocked AI agents from making purchases. Delta Air Lines has also officially announced that it does not currently offer a feature allowing external AI agents to book tickets on behalf of consumers.
As the industry prepares for the era of AI agents becoming businesses' new “customers,” demand is growing for urgently needed new internet access rules that can distinguish AI agents from malicious bots and verify their permissions.

[email protected] Lee Gu-sun Reporter