Meta has launched Muse, a personal artificial intelligence agent that can search for products, navigate online checkouts and prepare purchases for approval, as technology companies compete to persuade consumers to hand over more of the shopping process.
The service, introduced in the United States on Tuesday, 8 September, is aimed at adults and is available through iOS, Android and the Muse website. It can also carry out tasks beyond shopping, including arranging travel, sending emails and working with connected apps.
Unlike a conventional chatbot, Muse is designed to act on a user’s behalf. It can draw on information shared in conversations and connected services to make recommendations, browse retailers and complete a checkout using Stripe’s Link service, although the customer must approve the final purchase.
Meta says Muse can remember details mentioned only once, such as dietary requirements, and turn an Instagram recipe saved by a user into a grocery list. The company is presenting the system as an assistant that can continue working in the background and return when it requires approval or when circumstances change.
However, the launch comes as shoppers remain reluctant to allow artificial intelligence to make decisions with their money. Traffic from AI assistants and chatbots to retail websites rose sharply during the 2025 holiday season, but research suggests that consumers are more comfortable using the technology to compare products or find discounts than choosing what to buy.
A Vogue Business survey found that only 31 per cent of respondents would outsource shopping to an AI agent that understood their tastes and purchasing history. Just 24 per cent trusted AI-generated recommendations and summaries, while 72 per cent said they would not share card details and substantial proportions refused to provide browsing or location data.
More than half of those questioned had never used AI to shop for fashion or beauty, and only 2 per cent said it consistently understood their personal style. The findings point to a divide between using AI as a research tool and allowing it to become the buyer.
Meta’s shopping AI faces a test of trust
Muse’s potential advantage is the depth of information Meta already holds about its users, including what they watch, follow, save and share across services such as Instagram and Facebook. That data could make its recommendations more personal, but it also raises the stakes for a company with a long record of scrutiny over privacy.
Meta says users decide which applications Muse can access and what permissions it receives. They can disconnect services, ask the agent to forget information and opt out of having their interactions used to train Meta’s AI models. The company also says Muse conversations and data will not be shared with its advertising systems.
The service uses a dedicated virtual machine, known as Muse Secure VM, to store the agent and a user’s data. Meta says Muse cannot see passwords or payment details held in secure storage, while a separate security system must approve internet activity and sensitive actions such as sending an email or making a purchase. ([about.fb.com](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/amp/?utm_source=openai))
At checkout, Link generates a one-time card number so the user’s real card details are not exposed to the merchant or the agent. Meta says eligible purchases will also receive Link’s protections covering issues including damaged or lost items, price reductions and certain returns. ([about.fb.com](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/amp/?utm_source=openai))
The company plans to introduce a further privacy measure later this year, with a Confidential VM in which the entire virtual machine, including conversations and personal data, will be encrypted using a key held only by the user. Integration with Shop Pay and 1Password is also planned. ([about.fb.com](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/amp/?utm_source=openai))
Those assurances will be tested against Meta’s history of privacy controversies, including the $5 billion penalty imposed by the US Federal Trade Commission in 2019 over allegations concerning Facebook’s handling of user information. Muse asks consumers to grant access to a broader picture of their lives in return for more tailored suggestions.
Meta is entering a market that already includes shopping and personal agents from Amazon, Google, Walmart, Disney, Uber and OpenAI. Amazon introduced Alexa for Shopping in May, while Disney began testing a conversational shopping assistant in its store app the following month.
Uber Eats’ Cart Assistant can turn a written prompt or photograph of a handwritten shopping list into a grocery basket informed by previous orders. Mastercard has forecast that one in 10 shoppers will use a personal AI agent to buy products or services by 2030.
Meta’s challenge is therefore not simply to make Muse capable of completing transactions, but to persuade people that handing over control is worth it. Consumers may welcome help with repetitive orders or comparing thousands of products, while remaining unwilling to give up the browsing, discovery and personal choice that make shopping more than a chore.
